{ "schema_version": "2", "generator": "scrollback/cmd/export", "archive": { "as_of": "2026-08-17T07:36:29Z", "as_of_basis": "max(price_fact.valid_from, price_fact.valid_to, event.observed_at, model.last_seen, model_alias.last_seen) — the latest moment this archive READ a catalogue, over every layer that records one. event.occurred_at and availability.available_to are excluded: both are announced EFFECTIVE dates, they run months ahead of the page announcing them, and an as-of stamp in the future claims a reading that never happened.", "source_schema_version": "0006_benchmark_run_settings.sql", "license": { "name": "CC-BY-4.0", "url": "https://creativecommons.org/licenses/by/4.0/", "note": "CC-BY-4.0 covers this derived dataset. Upstream attribution is a condition: the measurements in benchmark_result derive entirely from Epoch AI's CC-BY-4.0 benchmark corpus (the descriptive benchmark and benchmark_version rows are this archive's own curated registry, not Epoch data); the price layer derives from the LiteLLM (MIT) and models.dev (MIT) catalogues, and NO price_fact row cites a vendor page; official vendor DEPRECATION pages back 44 availability rows and the 44 event rows derived from them. See DATA-LICENSE.", "attribution": { "vendor_deprecation_hosts": [ "developers.openai.com", "platform.claude.com" ], "announced_availability_rows": 44, "vendor_event_rows": 44, "evidence_rows_examined": 2343, "evidence_urls_present": 1245, "republished_hosts": [ { "host": "arcprize.org", "benchmark_result_rows": 336, "event_rows": 32 }, { "host": "htihle.github.io", "benchmark_result_rows": 318, "event_rows": 48 } ] } }, "determinism_note": "no wall clock enters this file: every timestamp comes out of the archive, so two runs over the same ledger produce byte-identical files (`make export-verify`).", "open_interval_note": "price_fact.valid_to IS NULL and availability.available_to IS NULL mean OPEN AT archive_as_of, not open today. A model with no availability row is UNKNOWN, not available." }, "rows": [ { "table": "model", "rows": 203 }, { "table": "model_alias", "rows": 3839 }, { "table": "offering", "rows": 1817 }, { "table": "price_fact", "rows": 6702 }, { "table": "benchmark", "rows": 13 }, { "table": "benchmark_version", "rows": 13 }, { "table": "benchmark_result", "rows": 2343 }, { "table": "availability", "rows": 1554 }, { "table": "event", "rows": 3588 }, { "table": "suspect", "rows": 31 } ], "ranking": { "default_job": "coding", "note": "the index ranks currently purchasable configurations on one named benchmark by task dollars per score point, lowest first. two efforts of the same model are two rows. a composite score is not computed.", "rule": "usd_per_point = cost_per_task / score; rank 1 is the lowest ratio among configurations that are not retired at the archive as-of. models with no availability window are unknown, not retired.", "jobs": [ { "id": "coding", "label": "Code", "benchmark": "ale-bench", "benchmark_name": "ALE-Bench", "metric": "rating", "score_kind": "rating", "measurer": "Epoch AI", "ratio_label": "$ / rating point", "ratio_note": "task cost divided by ALE-Bench rating — dollars per rating point", "cost_basis": "list_price_x_tokens", "cost_provenance": "community_submitted", "as_of": "2026-08-17T07:36:29Z", "rows": [ { "rank": 1, "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 566.05, "score_display": "566.05", "cost_usd_per_task": "0.000717", "cost_display": "0.000717", "usd_per_point": "0.0000012666725554279657", "usd_per_point_display": "0.000001267", "released_on": "2025-08-05" }, { "rank": 2, "model_uid": "openai/gpt-oss-120b", "display": "gpt-oss-120b", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 575.62, "score_display": "575.62", "cost_usd_per_task": "0.001303", "cost_display": "0.001303", "usd_per_point": "0.000002263646155449776", "usd_per_point_display": "0.000002264", "released_on": "2025-08-05" }, { "rank": 3, "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (none)", "config_key": "8f3301f5de45bc531c5ec6c881fa175b153ecff20c705f683c262d1c09067b64", "effort": "none", "score": 324.98, "score_display": "324.98", "cost_usd_per_task": "0.000882", "cost_display": "0.000882", "usd_per_point": "0.000002714013170041233", "usd_per_point_display": "0.000002714", "released_on": "2026-04-24" }, { "rank": 4, "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 678.2, "score_display": "678.2", "cost_usd_per_task": "0.005179", "cost_display": "0.005179", "usd_per_point": "0.000007636390445296372", "usd_per_point_display": "0.000007636", "released_on": "2026-04-24" }, { "rank": 5, "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 718.67, "score_display": "718.67", "cost_usd_per_task": "0.007194", "cost_display": "0.007194", "usd_per_point": "0.000010010157652329999", "usd_per_point_display": "0.00001001", "released_on": "2025-08-07" }, { "rank": 6, "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "config_key": "f3f3607ec6d855965dff8aaca3ae425e1c5322e2217c92b8434b5a84ed17d257", "effort": "max", "score": 1306.08, "score_display": "1306.08", "cost_usd_per_task": "0.019526", "cost_display": "0.019526", "usd_per_point": "0.000014950079627587899", "usd_per_point_display": "0.00001495", "released_on": "2026-07-31" }, { "rank": 7, "model_uid": "xai/grok-code-fast-1", "display": "Grok Code Fast 1", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 587.73, "score_display": "587.73", "cost_usd_per_task": "0.008947", "cost_display": "0.008947", "usd_per_point": "0.00001522297653684515", "usd_per_point_display": "0.00001522", "released_on": "2025-08-28" }, { "rank": 8, "model_uid": "mistral/mistral-large-2512", "display": "Mistral Large (2512)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 264.7, "score_display": "264.7", "cost_usd_per_task": "0.005104", "cost_display": "0.005104", "usd_per_point": "0.000019282206271250473", "usd_per_point_display": "0.00001928", "released_on": "2025-12-02" }, { "rank": 9, "model_uid": "mistral/codestral-2508", "display": "Codestral (2508)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 137.78, "score_display": "137.78", "cost_usd_per_task": "0.00278", "cost_display": "0.00278", "usd_per_point": "0.000020177093917840034", "usd_per_point_display": "0.00002018", "released_on": "2025-07-30" }, { "rank": 10, "model_uid": "xai/grok-4-1-fast-reasoning", "display": "Grok 4.1 Fast (reasoning)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 394.93, "score_display": "394.93", "cost_usd_per_task": "0.009221", "cost_display": "0.009221", "usd_per_point": "0.000023348441495961308", "usd_per_point_display": "0.00002335", "released_on": "2025-11-19" }, { "rank": 11, "model_uid": "deepseek/deepseek-v3.1-terminus", "display": "DeepSeek-V3.1-Terminus", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 745.17, "score_display": "745.17", "cost_usd_per_task": "0.01883", "cost_display": "0.01883", "usd_per_point": "0.00002526940161305474", "usd_per_point_display": "0.00002527", "released_on": "2025-09-22" }, { "rank": 12, "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 799.77, "score_display": "799.77", "cost_usd_per_task": "0.020554", "cost_display": "0.020554", "usd_per_point": "0.000025699888718006426", "usd_per_point_display": "0.0000257", "released_on": "2025-08-07" }, { "rank": 13, "model_uid": "openai/gpt-5.4-nano-20260317", "display": "GPT-5.4 nano (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1004.52, "score_display": "1004.52", "cost_usd_per_task": "0.027084", "cost_display": "0.027084", "usd_per_point": "0.000026962131167124597", "usd_per_point_display": "0.00002696", "released_on": "2026-03-17" }, { "rank": 14, "model_uid": "mistral/mistral-small-2603", "display": "Mistral Small (2603)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 497.62, "score_display": "497.62", "cost_usd_per_task": "0.015587", "cost_display": "0.015587", "usd_per_point": "0.000031323097946224026", "usd_per_point_display": "0.00003132", "released_on": "2026-03-16" }, { "rank": 15, "model_uid": "google/gemini-3.1-flash-lite", "display": "Gemini 3.1 Flash-Lite", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 797.73, "score_display": "797.73", "cost_usd_per_task": "0.025178", "cost_display": "0.025178", "usd_per_point": "0.00003156205733769571", "usd_per_point_display": "0.00003156", "released_on": "2026-03-03" }, { "rank": 16, "model_uid": "deepseek/deepseek-v4-pro", "display": "DeepSeek-V4 Pro (none)", "config_key": "8f3301f5de45bc531c5ec6c881fa175b153ecff20c705f683c262d1c09067b64", "effort": "none", "score": 521.67, "score_display": "521.67", "cost_usd_per_task": "0.0166", "cost_display": "0.0166", "usd_per_point": "0.00003182088293365538", "usd_per_point_display": "0.00003182", "released_on": "2026-04-24" }, { "rank": 17, "model_uid": "mistral/mistral-medium-2508", "display": "Mistral Medium 3.1 (2508)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 210.18, "score_display": "210.18", "cost_usd_per_task": "0.007372", "cost_display": "0.007372", "usd_per_point": "0.000035074697878009325", "usd_per_point_display": "0.00003507", "released_on": "2025-08-12" }, { "rank": 18, "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (none)", "config_key": "8f3301f5de45bc531c5ec6c881fa175b153ecff20c705f683c262d1c09067b64", "effort": "none", "score": 1086.03, "score_display": "1086.03", "cost_usd_per_task": "0.052314", "cost_display": "0.052314", "usd_per_point": "0.000048169940056904504", "usd_per_point_display": "0.00004817", "released_on": "2026-03-05" }, { "rank": 19, "model_uid": "openai/gpt-4.1-20250414", "display": "GPT-4.1", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 558.1, "score_display": "558.1", "cost_usd_per_task": "0.026917", "cost_display": "0.026917", "usd_per_point": "0.00004822970793764558", "usd_per_point_display": "0.00004823", "released_on": "2025-04-14" }, { "rank": 20, "model_uid": "openai/o4-mini-20250416", "display": "o4-mini (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 826.17, "score_display": "826.17", "cost_usd_per_task": "0.044719", "cost_display": "0.044719", "usd_per_point": "0.0000541280850188218", "usd_per_point_display": "0.00005413", "released_on": "2025-04-16" }, { "rank": 21, "model_uid": "openai/gpt-5-20250807", "display": "GPT-5 (minimal)", "config_key": "77c7caaaa0e1791a9e65655863014c4a23b8e5513a241f07be25e6c97ef3eea6", "effort": "minimal", "score": 807.65, "score_display": "807.65", "cost_usd_per_task": "0.043744", "cost_display": "0.043744", "usd_per_point": "0.000054162075156317714", "usd_per_point_display": "0.00005416", "released_on": "2025-08-07" }, { "rank": 22, "model_uid": "deepseek/deepseek-r1-0528", "display": "DeepSeek-R1 (May 2025)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 804.12, "score_display": "804.12", "cost_usd_per_task": "0.043794", "cost_display": "0.043794", "usd_per_point": "0.0000544620205939412", "usd_per_point_display": "0.00005446", "released_on": "2025-05-28" }, { "rank": 23, "model_uid": "xai/grok-4.3", "display": "Grok 4.3", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 944.17, "score_display": "944.17", "cost_usd_per_task": "0.055799", "cost_display": "0.055799", "usd_per_point": "0.00005909846743700817", "usd_per_point_display": "0.0000591", "released_on": "2026-04-17" }, { "rank": 24, "model_uid": "google/gemini-3.5-flash-lite", "display": "Gemini 3.5 Flash-Lite (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 765.27, "score_display": "765.27", "cost_usd_per_task": "0.045726", "cost_display": "0.045726", "usd_per_point": "0.000059751460268924696", "usd_per_point_display": "0.00005975", "released_on": "2026-07-21" }, { "rank": 25, "model_uid": "anthropic/claude-opus-4-6-20260205", "display": "Claude Opus 4.6", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 996.5, "score_display": "996.5", "cost_usd_per_task": "0.061688", "cost_display": "0.061688", "usd_per_point": "0.00006190466633216256", "usd_per_point_display": "0.0000619", "released_on": "2026-02-05" }, { "rank": 26, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (none)", "config_key": "8f3301f5de45bc531c5ec6c881fa175b153ecff20c705f683c262d1c09067b64", "effort": "none", "score": 1127.58, "score_display": "1127.58", "cost_usd_per_task": "0.073484", "cost_display": "0.073484", "usd_per_point": "0.00006516965536813353", "usd_per_point_display": "0.00006517", "released_on": "2026-04-23" }, { "rank": 27, "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1367.2, "score_display": "1367.2", "cost_usd_per_task": "0.091067", "cost_display": "0.091067", "usd_per_point": "0.00006660839672322995", "usd_per_point_display": "0.00006661", "released_on": "2025-12-17" }, { "rank": 28, "model_uid": "google/gemini-2.5-flash", "display": "Gemini 2.5 Flash", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 661.88, "score_display": "661.88", "cost_usd_per_task": "0.068155", "cost_display": "0.068155", "usd_per_point": "0.00010297183779537076", "usd_per_point_display": "0.000103", "released_on": "2025-06-17" }, { "rank": 29, "model_uid": "anthropic/claude-haiku-4-5-20251001", "display": "Claude Haiku 4.5", "config_key": "f85e697f5a4a9f083db84d48bb84c0196a2558fbcddebfd6607318bfe3d05ebf", "effort": "", "score": 653.48, "score_display": "653.48", "cost_usd_per_task": "0.071831", "cost_display": "0.071831", "usd_per_point": "0.00010992073208055336", "usd_per_point_display": "0.0001099", "released_on": "2025-10-15" }, { "rank": 30, "model_uid": "openai/o3-20250416", "display": "o3 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 933.55, "score_display": "933.55", "cost_usd_per_task": "0.103474", "cost_display": "0.103474", "usd_per_point": "0.00011083926945530502", "usd_per_point_display": "0.0001108", "released_on": "2025-04-16" }, { "rank": 31, "model_uid": "openai/gpt-5.2-20251211", "display": "GPT-5.2 (medium)", "config_key": "e79d13ddd0be01bb677c65063551a6652338b0321ce1dd6ffb60d0f2c086d8f4", "effort": "medium", "score": 1249.83, "score_display": "1249.83", "cost_usd_per_task": "0.142947", "cost_display": "0.142947", "usd_per_point": "0.00011437315474904586", "usd_per_point_display": "0.0001144", "released_on": "2025-12-11" }, { "rank": 32, "model_uid": "xai/grok-4.20-0309-reasoning", "display": "Grok 4.20 (reasoning)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1150.28, "score_display": "1150.28", "cost_usd_per_task": "0.13838", "cost_display": "0.13838", "usd_per_point": "0.00012030114406927009", "usd_per_point_display": "0.0001203", "released_on": "2026-02-17" }, { "rank": 33, "model_uid": "xai/grok-4.5", "display": "Grok 4.5 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1309.28, "score_display": "1309.28", "cost_usd_per_task": "0.176767", "cost_display": "0.176767", "usd_per_point": "0.00013501084565562752", "usd_per_point_display": "0.000135", "released_on": "2026-07-08" }, { "rank": 34, "model_uid": "deepseek/deepseek-v4-pro", "display": "DeepSeek-V4 Pro (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1006.08, "score_display": "1006.08", "cost_usd_per_task": "0.148062", "cost_display": "0.148062", "usd_per_point": "0.00014716722328244274", "usd_per_point_display": "0.0001472", "released_on": "2026-04-24" }, { "rank": 35, "model_uid": "anthropic/claude-opus-4-7-20260416", "display": "Claude Opus 4.7", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1323.05, "score_display": "1323.05", "cost_usd_per_task": "0.224549", "cost_display": "0.224549", "usd_per_point": "0.0001697207210611844", "usd_per_point_display": "0.0001697", "released_on": "2026-04-16" }, { "rank": 36, "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (medium)", "config_key": "e79d13ddd0be01bb677c65063551a6652338b0321ce1dd6ffb60d0f2c086d8f4", "effort": "medium", "score": 1520.72, "score_display": "1520.72", "cost_usd_per_task": "0.26258", "cost_display": "0.26258", "usd_per_point": "0.00017266820979536007", "usd_per_point_display": "0.0001727", "released_on": "2026-03-05" }, { "rank": 37, "model_uid": "anthropic/claude-sonnet-4-5-20250929", "display": "Claude Sonnet 4.5", "config_key": "f85e697f5a4a9f083db84d48bb84c0196a2558fbcddebfd6607318bfe3d05ebf", "effort": "", "score": 796.15, "score_display": "796.15", "cost_usd_per_task": "0.151558", "cost_display": "0.151558", "usd_per_point": "0.00019036362494504806", "usd_per_point_display": "0.0001904", "released_on": "2025-09-29" }, { "rank": 38, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (medium)", "config_key": "e79d13ddd0be01bb677c65063551a6652338b0321ce1dd6ffb60d0f2c086d8f4", "effort": "medium", "score": 1589.38, "score_display": "1589.38", "cost_usd_per_task": "0.328965", "cost_display": "0.328965", "usd_per_point": "0.00020697693440209389", "usd_per_point_display": "0.000207", "released_on": "2026-04-23" }, { "rank": 39, "model_uid": "openai/gpt-5.1-codex-max", "display": "GPT-5.1-Codex-Max", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1208.83, "score_display": "1208.83", "cost_usd_per_task": "0.253186", "cost_display": "0.253186", "usd_per_point": "0.00020944715137777853", "usd_per_point_display": "0.0002094", "released_on": "2025-11-19" }, { "rank": 40, "model_uid": "openai/gpt-5-20250807", "display": "GPT-5 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1162.45, "score_display": "1162.45", "cost_usd_per_task": "0.244948", "cost_display": "0.244948", "usd_per_point": "0.00021071702008688544", "usd_per_point_display": "0.0002107", "released_on": "2025-08-07" }, { "rank": 41, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "config_key": "f3f3607ec6d855965dff8aaca3ae425e1c5322e2217c92b8434b5a84ed17d257", "effort": "max", "score": 1667.4, "score_display": "1667.4", "cost_usd_per_task": "0.359171", "cost_display": "0.359171", "usd_per_point": "0.00021540782055895407", "usd_per_point_display": "0.0002154", "released_on": "2026-07-09" }, { "rank": 42, "model_uid": "google/gemini-3-pro-preview", "display": "Gemini 3 Pro Preview", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1176.75, "score_display": "1176.75", "cost_usd_per_task": "0.259385", "cost_display": "0.259385", "usd_per_point": "0.00022042489908646694", "usd_per_point_display": "0.0002204", "released_on": "2025-11-18" }, { "rank": 43, "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1188.58, "score_display": "1188.58", "cost_usd_per_task": "0.271408", "cost_display": "0.271408", "usd_per_point": "0.00022834643019401302", "usd_per_point_display": "0.0002283", "released_on": "2026-03-17" }, { "rank": 44, "model_uid": "openai/gpt-5.2-20251211", "display": "GPT-5.2 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1293.55, "score_display": "1293.55", "cost_usd_per_task": "0.296865", "cost_display": "0.296865", "usd_per_point": "0.0002294963472614124", "usd_per_point_display": "0.0002295", "released_on": "2025-12-11" }, { "rank": 45, "model_uid": "openai/gpt-5.1-20251113", "display": "GPT-5.1 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1192.15, "score_display": "1192.15", "cost_usd_per_task": "0.315236", "cost_display": "0.315236", "usd_per_point": "0.00026442645640229834", "usd_per_point_display": "0.0002644", "released_on": "2025-11-13" }, { "rank": 46, "model_uid": "anthropic/claude-sonnet-4-20250514", "display": "Claude Sonnet 4", "config_key": "f85e697f5a4a9f083db84d48bb84c0196a2558fbcddebfd6607318bfe3d05ebf", "effort": "", "score": 655.35, "score_display": "655.35", "cost_usd_per_task": "0.185869", "cost_display": "0.185869", "usd_per_point": "0.00028361791409170674", "usd_per_point_display": "0.0002836", "released_on": "2025-05-22" }, { "rank": 47, "model_uid": "anthropic/claude-opus-4-5-20251101", "display": "Claude Opus 4.5", "config_key": "bd74c0a505f8dcecf84b54068b7f688894bf0b5d11bacbe1552ca89d2e65fbfd", "effort": "", "score": 1025.38, "score_display": "1025.38", "cost_usd_per_task": "0.292547", "cost_display": "0.292547", "usd_per_point": "0.00028530593536054923", "usd_per_point_display": "0.0002853", "released_on": "2025-11-24" }, { "rank": 48, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (none)", "config_key": "8f3301f5de45bc531c5ec6c881fa175b153ecff20c705f683c262d1c09067b64", "effort": "none", "score": 1411.83, "score_display": "1411.83", "cost_usd_per_task": "0.421746", "cost_display": "0.421746", "usd_per_point": "0.00029872293406429956", "usd_per_point_display": "0.0002987", "released_on": "2026-05-28" }, { "rank": 49, "model_uid": "google/gemini-3.1-pro-preview", "display": "Gemini 3.1 Pro Preview", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1160.6, "score_display": "1160.6", "cost_usd_per_task": "0.388403", "cost_display": "0.388403", "usd_per_point": "0.00033465707392727904", "usd_per_point_display": "0.0003347", "released_on": "2026-02-19" }, { "rank": 50, "model_uid": "openai/gpt-5.3-codex", "display": "GPT-5.3-Codex (xhigh)", "config_key": "82773910ecd1b341391c6ed673ce8cdec8c8ef280116e45ac34f4cedf54b9558", "effort": "xhigh", "score": 1655.22, "score_display": "1655.22", "cost_usd_per_task": "0.559265", "cost_display": "0.559265", "usd_per_point": "0.00033787955679607547", "usd_per_point_display": "0.0003379", "released_on": "2026-02-05" }, { "rank": 51, "model_uid": "google/gemini-3.6-flash", "display": "Gemini 3.6 Flash (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 715.52, "score_display": "715.52", "cost_usd_per_task": "0.244468", "cost_display": "0.244468", "usd_per_point": "0.00034166480322003577", "usd_per_point_display": "0.0003417", "released_on": "2026-07-21" }, { "rank": 52, "model_uid": "google/gemini-2.5-pro", "display": "Gemini 2.5 Pro", "config_key": "f85e697f5a4a9f083db84d48bb84c0196a2558fbcddebfd6607318bfe3d05ebf", "effort": "", "score": 785.52, "score_display": "785.52", "cost_usd_per_task": "0.26988", "cost_display": "0.26988", "usd_per_point": "0.0003435685915062634", "usd_per_point_display": "0.0003436", "released_on": "2025-06-17" }, { "rank": 53, "model_uid": "mistral/mistral-medium-2604", "display": "Mistral Medium (2604)", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 763.98, "score_display": "763.98", "cost_usd_per_task": "0.272986", "cost_display": "0.272986", "usd_per_point": "0.0003573208722741433", "usd_per_point_display": "0.0003573", "released_on": "2026-04-28" }, { "rank": 54, "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1607, "score_display": "1607", "cost_usd_per_task": "0.612174", "cost_display": "0.612174", "usd_per_point": "0.0003809421281891724", "usd_per_point_display": "0.0003809", "released_on": "2026-03-05" }, { "rank": 55, "model_uid": "openai/gpt-5.1-codex", "display": "GPT-5.1-Codex", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1244.92, "score_display": "1244.92", "cost_usd_per_task": "0.512093", "cost_display": "0.512093", "usd_per_point": "0.00041134611059345176", "usd_per_point_display": "0.0004113", "released_on": "2025-11-12" }, { "rank": 56, "model_uid": "openai/gpt-5.2-codex", "display": "GPT-5.2-Codex", "config_key": "e51dfe78986ee35c325753e7c0f8487bd6ab5d19b23c242583b1a74b2a591bf6", "effort": "", "score": 1299.9, "score_display": "1299.9", "cost_usd_per_task": "0.587753", "cost_display": "0.587753", "usd_per_point": "0.00045215247326717436", "usd_per_point_display": "0.0004522", "released_on": "2025-12-18" }, { "rank": 57, "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 911.02, "score_display": "911.02", "cost_usd_per_task": "0.441921", "cost_display": "0.441921", "usd_per_point": "0.0004850837522776668", "usd_per_point_display": "0.0004851", "released_on": "2026-05-19" }, { "rank": 58, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1563.83, "score_display": "1563.83", "cost_usd_per_task": "0.963945", "cost_display": "0.963945", "usd_per_point": "0.0006164001202176708", "usd_per_point_display": "0.0006164", "released_on": "2026-05-28" }, { "rank": 59, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (max)", "config_key": "f3f3607ec6d855965dff8aaca3ae425e1c5322e2217c92b8434b5a84ed17d257", "effort": "max", "score": 1951.42, "score_display": "1951.42", "cost_usd_per_task": "1.254047", "cost_display": "1.25405", "usd_per_point": "0.0006426330569533979", "usd_per_point_display": "0.0006426", "released_on": "2026-07-09" }, { "rank": 60, "model_uid": "anthropic/claude-opus-4-1-20250805", "display": "Claude Opus 4.1", "config_key": "bd74c0a505f8dcecf84b54068b7f688894bf0b5d11bacbe1552ca89d2e65fbfd", "effort": "", "score": 674.77, "score_display": "674.77", "cost_usd_per_task": "0.434347", "cost_display": "0.434347", "usd_per_point": "0.0006436963706151726", "usd_per_point_display": "0.0006437", "released_on": "2025-08-05" }, { "rank": 61, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 1463.12, "score_display": "1463.12", "cost_usd_per_task": "0.96046", "cost_display": "0.96046", "usd_per_point": "0.0006564464978949095", "usd_per_point_display": "0.0006564", "released_on": "2026-06-30" }, { "rank": 62, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "config_key": "f3f3607ec6d855965dff8aaca3ae425e1c5322e2217c92b8434b5a84ed17d257", "effort": "max", "score": 2176.88, "score_display": "2176.88", "cost_usd_per_task": "1.534078", "cost_display": "1.53408", "usd_per_point": "0.0007047140862151336", "usd_per_point_display": "0.0007047", "released_on": "2026-07-09" }, { "rank": 63, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (xhigh)", "config_key": "82773910ecd1b341391c6ed673ce8cdec8c8ef280116e45ac34f4cedf54b9558", "effort": "xhigh", "score": 1942.97, "score_display": "1942.97", "cost_usd_per_task": "1.517132", "cost_display": "1.51713", "usd_per_point": "0.0007808314075873534", "usd_per_point_display": "0.0007808", "released_on": "2026-04-23" }, { "rank": 64, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 2164.57, "score_display": "2164.57", "cost_usd_per_task": "1.890512", "cost_display": "1.89051", "usd_per_point": "0.00087338917198335", "usd_per_point_display": "0.0008734", "released_on": "2026-07-24" }, { "rank": 65, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (high)", "config_key": "f201028ac46b1f5e40d0b4cac5e073109d112421fa541ca1d3f947b65bdc19fb", "effort": "high", "score": 2041.3, "score_display": "2041.3", "cost_usd_per_task": "2.108822", "cost_display": "2.10882", "usd_per_point": "0.001033077940528095", "usd_per_point_display": "0.001033", "released_on": "2026-06-09" }, { "rank": 66, "model_uid": "anthropic/claude-sonnet-4-6", "display": "Claude Sonnet 4.6 (medium)", "config_key": "e79d13ddd0be01bb677c65063551a6652338b0321ce1dd6ffb60d0f2c086d8f4", "effort": "medium", "score": 1327.3, "score_display": "1327.3", "cost_usd_per_task": "1.770753", "cost_display": "1.77075", "usd_per_point": "0.0013341015595569954", "usd_per_point_display": "0.001334", "released_on": "2026-02-17" } ], "excluded_unbuyable": 0, "excluded_no_cost": 0, "excluded_zero_cost": 0, "excluded_nonpositive_score": 0, "note": "66 ranked, 0 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score" }, { "id": "reasoning", "label": "Reason", "benchmark": "proofbench", "benchmark_name": "ProofBench", "metric": "accuracy", "score_kind": "fraction", "measurer": "Epoch AI", "ratio_label": "$ / correct answer", "ratio_note": "task cost divided by accuracy — expected spend per correct answer at this accuracy", "cost_basis": "unknown", "cost_provenance": "leaderboard_maintainer_verified", "as_of": "2026-08-17T07:36:29Z", "rows": [ { "rank": 1, "model_uid": "openai/gpt-5.4-nano-20260317", "display": "GPT-5.4 nano (high)", "config_key": "81e45dfc9b17a4b37ab7d1d92c691d69881f3c68f7e1db31fc9c422dbbd389e3", "effort": "high", "score": 0.05, "score_display": "0.05", "cost_usd_per_task": "0.015894", "cost_display": "0.015894", "usd_per_point": "0.31788", "usd_per_point_display": "0.3179", "released_on": "2026-03-17" }, { "rank": 2, "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "config_key": "81e45dfc9b17a4b37ab7d1d92c691d69881f3c68f7e1db31fc9c422dbbd389e3", "effort": "high", "score": 0.12, "score_display": "0.12", "cost_usd_per_task": "0.049218", "cost_display": "0.049218", "usd_per_point": "0.41015", "usd_per_point_display": "0.4102", "released_on": "2025-08-07" }, { "rank": 3, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "config_key": "ccf092edbfd284b1ba2d6880080e2310ab5b2cbc605b406fc640920f0f29fe1d", "effort": "max", "score": 0.54, "score_display": "0.54", "cost_usd_per_task": "0.449325", "cost_display": "0.449325", "usd_per_point": "0.8320833333333333", "usd_per_point_display": "0.8321", "released_on": "2026-07-09" }, { "rank": 4, "model_uid": "google/gemini-3.5-flash-lite", "display": "Gemini 3.5 Flash-Lite (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.13, "score_display": "0.13", "cost_usd_per_task": "0.132788", "cost_display": "0.132788", "usd_per_point": "1.0214461538461537", "usd_per_point_display": "1.021", "released_on": "2026-07-21" }, { "rank": 5, "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "config_key": "0284a691fb397a0e1509a31982ab9699a31d1a70219d87585a24d80ffec9d72e", "effort": "xhigh", "score": 0.21, "score_display": "0.21", "cost_usd_per_task": "0.249441", "cost_display": "0.249441", "usd_per_point": "1.1878142857142857", "usd_per_point_display": "1.188", "released_on": "2026-03-17" }, { "rank": 6, "model_uid": "google/gemini-3.6-flash", "display": "Gemini 3.6 Flash (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.36, "score_display": "0.36", "cost_usd_per_task": "0.522059", "cost_display": "0.522059", "usd_per_point": "1.450163888888889", "usd_per_point_display": "1.45", "released_on": "2026-07-21" }, { "rank": 7, "model_uid": "deepseek/deepseek-v3.2", "display": "DeepSeek-V3.2 (none)", "config_key": "e12f53954fc4190d0f1c5a4eff26678decda0ec9282510e254d6f1ed1a3b5ec2", "effort": "none", "score": 0.08, "score_display": "0.08", "cost_usd_per_task": "0.11773", "cost_display": "0.11773", "usd_per_point": "1.471625", "usd_per_point_display": "1.472", "released_on": "2025-12-01" }, { "rank": 8, "model_uid": "xai/grok-4.5", "display": "Grok 4.5 (high)", "config_key": "472a7c4c5d6c196208b83c00f16b3c800290bfbf6fcbc747ad3a96d64ffb0838", "effort": "high", "score": 0.3, "score_display": "0.3", "cost_usd_per_task": "0.484548", "cost_display": "0.484548", "usd_per_point": "1.61516", "usd_per_point_display": "1.615", "released_on": "2026-07-08" }, { "rank": 9, "model_uid": "google/gemini-3.1-pro-preview", "display": "Gemini 3.1 Pro Preview (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.26, "score_display": "0.26", "cost_usd_per_task": "0.525953", "cost_display": "0.525953", "usd_per_point": "2.022896153846154", "usd_per_point_display": "2.023", "released_on": "2026-02-19" }, { "rank": 10, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.77, "score_display": "0.77", "cost_usd_per_task": "1.635455", "cost_display": "1.63546", "usd_per_point": "2.1239675324675327", "usd_per_point_display": "2.124", "released_on": "2026-06-09" }, { "rank": 11, "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.15, "score_display": "0.15", "cost_usd_per_task": "0.327286", "cost_display": "0.327286", "usd_per_point": "2.181906666666667", "usd_per_point_display": "2.182", "released_on": "2025-12-17" }, { "rank": 12, "model_uid": "xai/grok-4.3", "display": "Grok 4.3 (high)", "config_key": "472a7c4c5d6c196208b83c00f16b3c800290bfbf6fcbc747ad3a96d64ffb0838", "effort": "high", "score": 0.11, "score_display": "0.11", "cost_usd_per_task": "0.244122", "cost_display": "0.244122", "usd_per_point": "2.219290909090909", "usd_per_point_display": "2.219", "released_on": "2026-04-17" }, { "rank": 13, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "config_key": "ccf092edbfd284b1ba2d6880080e2310ab5b2cbc605b406fc640920f0f29fe1d", "effort": "max", "score": 0.77, "score_display": "0.77", "cost_usd_per_task": "1.736601", "cost_display": "1.7366", "usd_per_point": "2.255325974025974", "usd_per_point_display": "2.255", "released_on": "2026-07-09" }, { "rank": 14, "model_uid": "xai/grok-4.20-0309-reasoning", "display": "Grok 4.20 (reasoning)", "config_key": "13bfad10661cd4d664635e61961c0df54ce4bff023d0a88690e4ad068972298a", "effort": "", "score": 0.14, "score_display": "0.14", "cost_usd_per_task": "0.31815", "cost_display": "0.31815", "usd_per_point": "2.2725", "usd_per_point_display": "2.272", "released_on": "2026-02-17" }, { "rank": 15, "model_uid": "xai/grok-4-1-fast-reasoning", "display": "Grok 4.1 Fast (reasoning)", "config_key": "13bfad10661cd4d664635e61961c0df54ce4bff023d0a88690e4ad068972298a", "effort": "", "score": 0.04, "score_display": "0.04", "cost_usd_per_task": "0.096578", "cost_display": "0.096578", "usd_per_point": "2.41445", "usd_per_point_display": "2.414", "released_on": "2025-11-19" }, { "rank": 16, "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (high)", "config_key": "81e45dfc9b17a4b37ab7d1d92c691d69881f3c68f7e1db31fc9c422dbbd389e3", "effort": "high", "score": 0.09, "score_display": "0.09", "cost_usd_per_task": "0.22021", "cost_display": "0.22021", "usd_per_point": "2.4467777777777777", "usd_per_point_display": "2.447", "released_on": "2025-08-07" }, { "rank": 17, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "config_key": "386226d7bea1488653c56e9292130b8099165d14b4267bffba5988ac77fd266e", "effort": "max", "score": 0.78, "score_display": "0.78", "cost_usd_per_task": "2.10619", "cost_display": "2.10619", "usd_per_point": "2.7002435897435895", "usd_per_point_display": "2.7", "released_on": "2026-07-24" }, { "rank": 18, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.6900000000000001, "score_display": "0.69", "cost_usd_per_task": "2.293129", "cost_display": "2.29313", "usd_per_point": "3.3233753623188402", "usd_per_point_display": "3.323", "released_on": "2026-05-28" }, { "rank": 19, "model_uid": "anthropic/claude-opus-4-7-20260416", "display": "Claude Opus 4.7 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.54, "score_display": "0.54", "cost_usd_per_task": "1.807786", "cost_display": "1.80779", "usd_per_point": "3.3477518518518514", "usd_per_point_display": "3.348", "released_on": "2026-04-16" }, { "rank": 20, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (xhigh)", "config_key": "0284a691fb397a0e1509a31982ab9699a31d1a70219d87585a24d80ffec9d72e", "effort": "xhigh", "score": 0.5, "score_display": "0.5", "cost_usd_per_task": "1.68179", "cost_display": "1.68179", "usd_per_point": "3.36358", "usd_per_point_display": "3.364", "released_on": "2026-04-23" }, { "rank": 21, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.66, "score_display": "0.66", "cost_usd_per_task": "2.248759", "cost_display": "2.24876", "usd_per_point": "3.407210606060606", "usd_per_point_display": "3.407", "released_on": "2026-06-30" }, { "rank": 22, "model_uid": "anthropic/claude-sonnet-4-5-20250929", "display": "Claude Sonnet 4.5 (unknown)", "config_key": "fb68d2661419f17085592634566b636449d0fb6a92e62d4407889b0fd072223b", "effort": "unknown", "score": 0.19, "score_display": "0.19", "cost_usd_per_task": "0.65341", "cost_display": "0.65341", "usd_per_point": "3.439", "usd_per_point_display": "3.439", "released_on": "2025-09-29" }, { "rank": 23, "model_uid": "anthropic/claude-opus-4-5-20251101", "display": "Claude Opus 4.5 (high)", "config_key": "670b54068ba2f1f204ac79527e3bce8bd5ba1b709ca12b484aecffcf33177cf8", "effort": "high", "score": 0.36, "score_display": "0.36", "cost_usd_per_task": "1.267043", "cost_display": "1.26704", "usd_per_point": "3.519563888888889", "usd_per_point_display": "3.52", "released_on": "2025-11-24" }, { "rank": 24, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "config_key": "0284a691fb397a0e1509a31982ab9699a31d1a70219d87585a24d80ffec9d72e", "effort": "xhigh", "score": 0.71, "score_display": "0.71", "cost_usd_per_task": "2.918608", "cost_display": "2.91861", "usd_per_point": "4.1107154929577465", "usd_per_point_display": "4.111", "released_on": "2026-07-09" }, { "rank": 25, "model_uid": "openai/gpt-5-20250807", "display": "GPT-5 (high)", "config_key": "81e45dfc9b17a4b37ab7d1d92c691d69881f3c68f7e1db31fc9c422dbbd389e3", "effort": "high", "score": 0.18, "score_display": "0.18", "cost_usd_per_task": "0.852361", "cost_display": "0.852361", "usd_per_point": "4.735338888888889", "usd_per_point_display": "4.735", "released_on": "2025-08-07" }, { "rank": 26, "model_uid": "anthropic/claude-sonnet-4-6", "display": "Claude Sonnet 4.6 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.45, "score_display": "0.45", "cost_usd_per_task": "2.162108", "cost_display": "2.16211", "usd_per_point": "4.804684444444444", "usd_per_point_display": "4.805", "released_on": "2026-02-17" }, { "rank": 27, "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.29, "score_display": "0.29", "cost_usd_per_task": "1.399975", "cost_display": "1.39997", "usd_per_point": "4.8275", "usd_per_point_display": "4.828", "released_on": "2026-05-19" }, { "rank": 28, "model_uid": "anthropic/claude-opus-4-6-20260205", "display": "Claude Opus 4.6 (max)", "config_key": "a7243fadf1ad37d6f9c5153e4598a9ad089ff703b36e66c280e07fd919d66210", "effort": "max", "score": 0.5, "score_display": "0.5", "cost_usd_per_task": "2.703583", "cost_display": "2.70358", "usd_per_point": "5.407166", "usd_per_point_display": "5.407", "released_on": "2026-02-05" }, { "rank": 29, "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (xhigh)", "config_key": "0284a691fb397a0e1509a31982ab9699a31d1a70219d87585a24d80ffec9d72e", "effort": "xhigh", "score": 0.56, "score_display": "0.56", "cost_usd_per_task": "3.202015", "cost_display": "3.20201", "usd_per_point": "5.717883928571427", "usd_per_point_display": "5.718", "released_on": "2026-03-05" }, { "rank": 30, "model_uid": "openai/gpt-5.1-codex-max", "display": "GPT-5.1-Codex-Max (high)", "config_key": "81e45dfc9b17a4b37ab7d1d92c691d69881f3c68f7e1db31fc9c422dbbd389e3", "effort": "high", "score": 0.09, "score_display": "0.09", "cost_usd_per_task": "0.527356", "cost_display": "0.527356", "usd_per_point": "5.859511111111112", "usd_per_point_display": "5.86", "released_on": "2025-11-19" }, { "rank": 31, "model_uid": "google/gemini-3-pro-preview", "display": "Gemini 3 Pro Preview (high)", "config_key": "6b1f25b544c2e74b17d1a39b68c48b00d8f715abe795ce73bbcca590de234888", "effort": "high", "score": 0.2, "score_display": "0.2", "cost_usd_per_task": "1.229076", "cost_display": "1.22908", "usd_per_point": "6.14538", "usd_per_point_display": "6.145", "released_on": "2025-11-18" }, { "rank": 32, "model_uid": "openai/gpt-5.2-20251211", "display": "GPT-5.2 (xhigh)", "config_key": "0284a691fb397a0e1509a31982ab9699a31d1a70219d87585a24d80ffec9d72e", "effort": "xhigh", "score": 0.15, "score_display": "0.15", "cost_usd_per_task": "1.059689", "cost_display": "1.05969", "usd_per_point": "7.064593333333335", "usd_per_point_display": "7.065", "released_on": "2025-12-11" }, { "rank": 33, "model_uid": "mistral/mistral-medium-2604", "display": "Mistral Medium (2604) (high)", "config_key": "12f542760b310772eda8817c9ec09208a0f6750be399dcd3c38b4c7b306e1ba7", "effort": "high", "score": 0.1, "score_display": "0.1", "cost_usd_per_task": "1.501427", "cost_display": "1.50143", "usd_per_point": "15.01427", "usd_per_point_display": "15.01", "released_on": "2026-04-28" } ], "excluded_unbuyable": 0, "excluded_no_cost": 0, "excluded_zero_cost": 1, "excluded_nonpositive_score": 0, "note": "33 ranked, 0 retired excluded, 0 dropped for missing cost, 1 for a published cost of $0, 0 for a non-positive score" }, { "id": "agent", "label": "Agent", "benchmark": "deepswe", "benchmark_name": "DeepSWE", "metric": "pass@1", "score_kind": "fraction", "measurer": "Epoch AI", "ratio_label": "$ / solved task", "ratio_note": "task cost divided by pass@1 — expected spend per solved task at this pass rate", "cost_basis": "unknown", "cost_provenance": "leaderboard_maintainer_verified", "as_of": "2026-08-17T07:36:29Z", "rows": [ { "rank": 1, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.3511111111111111, "score_display": "0.351", "cost_usd_per_task": "0.5832493688888889", "cost_display": "0.583249", "usd_per_point": "1.661153265822785", "usd_per_point_display": "1.661", "released_on": "2026-07-09" }, { "rank": 2, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.4424778761061947, "score_display": "0.442", "cost_usd_per_task": "0.7779007499999999", "cost_display": "0.777901", "usd_per_point": "1.7580556949999997", "usd_per_point_display": "1.758", "released_on": "2026-07-09" }, { "rank": 3, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.24053452115812918, "score_display": "0.241", "cost_usd_per_task": "0.4277452605790646", "cost_display": "0.427745", "usd_per_point": "1.7783113148148149", "usd_per_point_display": "1.778", "released_on": "2026-07-09" }, { "rank": 4, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.11283185840707964, "score_display": "0.113", "cost_usd_per_task": "0.21630978982300886", "cost_display": "0.21631", "usd_per_point": "1.9170985294117648", "usd_per_point_display": "1.917", "released_on": "2026-07-09" }, { "rank": 5, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.5376106194690266, "score_display": "0.538", "cost_usd_per_task": "1.1343762765486725", "cost_display": "1.13438", "usd_per_point": "2.1100332386831275", "usd_per_point_display": "2.11", "released_on": "2026-07-09" }, { "rank": 6, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.45353982300884954, "score_display": "0.454", "cost_usd_per_task": "1.0742664646017699", "cost_display": "1.07427", "usd_per_point": "2.3686265463414635", "usd_per_point_display": "2.369", "released_on": "2026-07-09" }, { "rank": 7, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.5685840707964602, "score_display": "0.569", "cost_usd_per_task": "1.5356192685840708", "cost_display": "1.53562", "usd_per_point": "2.7007778575875485", "usd_per_point_display": "2.701", "released_on": "2026-07-09" }, { "rank": 8, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.5812917594654788, "score_display": "0.581", "cost_usd_per_task": "1.6626223919821828", "cost_display": "1.66262", "usd_per_point": "2.8602201302681993", "usd_per_point_display": "2.86", "released_on": "2026-07-24" }, { "rank": 9, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.6106194690265486, "score_display": "0.611", "cost_usd_per_task": "1.8620318075221238", "cost_display": "1.86203", "usd_per_point": "3.04941440942029", "usd_per_point_display": "3.049", "released_on": "2026-07-09" }, { "rank": 10, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.6017699115044248, "score_display": "0.602", "cost_usd_per_task": "2.127191725663717", "cost_display": "2.12719", "usd_per_point": "3.534892132352941", "usd_per_point_display": "3.535", "released_on": "2026-07-09" }, { "rank": 11, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.26991150442477874, "score_display": "0.27", "cost_usd_per_task": "1.2002025442477877", "cost_display": "1.2002", "usd_per_point": "4.4466520491803285", "usd_per_point_display": "4.447", "released_on": "2026-04-23" }, { "rank": 12, "model_uid": "xai/grok-4.5", "display": "Grok 4.5 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.5376106194690266, "score_display": "0.538", "cost_usd_per_task": "2.4157481725663716", "cost_display": "2.41575", "usd_per_point": "4.493490427983539", "usd_per_point_display": "4.493", "released_on": "2026-07-08" }, { "rank": 13, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.671875, "score_display": "0.672", "cost_usd_per_task": "3.028116810267857", "cost_display": "3.02812", "usd_per_point": "4.506964554817276", "usd_per_point_display": "4.507", "released_on": "2026-07-09" }, { "rank": 14, "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.015486725663716814, "score_display": "0.0155", "cost_usd_per_task": "0.07240623938053097", "cost_display": "0.0724062", "usd_per_point": "4.675374314285714", "usd_per_point_display": "4.675", "released_on": "2026-07-09" }, { "rank": 15, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.6890380313199106, "score_display": "0.689", "cost_usd_per_task": "3.28978764541387", "cost_display": "3.28979", "usd_per_point": "4.7744645373376615", "usd_per_point_display": "4.774", "released_on": "2026-07-24" }, { "rank": 16, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.6940133037694013, "score_display": "0.694", "cost_usd_per_task": "3.4698331108647453", "cost_display": "3.46983", "usd_per_point": "4.999663683706071", "usd_per_point_display": "5", "released_on": "2026-07-09" }, { "rank": 17, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.5398230088495575, "score_display": "0.54", "cost_usd_per_task": "2.749342734513274", "cost_display": "2.74934", "usd_per_point": "5.093044737704918", "usd_per_point_display": "5.093", "released_on": "2026-04-23" }, { "rank": 18, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.4079822616407982, "score_display": "0.408", "cost_usd_per_task": "2.293369216740577", "cost_display": "2.29337", "usd_per_point": "5.621247373641306", "usd_per_point_display": "5.621", "released_on": "2026-05-28" }, { "rank": 19, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.5958429561200924, "score_display": "0.596", "cost_usd_per_task": "3.757873125874126", "cost_display": "3.75787", "usd_per_point": "6.306818075594947", "usd_per_point_display": "6.307", "released_on": "2026-06-09" }, { "rank": 20, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.7073170731707318, "score_display": "0.707", "cost_usd_per_task": "4.703663860310422", "cost_display": "4.70366", "usd_per_point": "6.650007526645768", "usd_per_point_display": "6.65", "released_on": "2026-07-09" }, { "rank": 21, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.48672566371681414, "score_display": "0.487", "cost_usd_per_task": "3.4438571028761062", "cost_display": "3.44386", "usd_per_point": "7.075560956818182", "usd_per_point_display": "7.076", "released_on": "2026-05-28" }, { "rank": 22, "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.6962305986696231, "score_display": "0.696", "cost_usd_per_task": "4.945847263858093", "cost_display": "4.94585", "usd_per_point": "7.103748777070063", "usd_per_point_display": "7.104", "released_on": "2026-07-09" }, { "rank": 23, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (low)", "config_key": "cd068b6bb9240e7ca929cf7ed557d6b13a6ed397c52dc0f5f1f84cf8555d92e8", "effort": "low", "score": 0.3051224944320713, "score_display": "0.305", "cost_usd_per_task": "2.1865550832962137", "cost_display": "2.18656", "usd_per_point": "7.166154981021897", "usd_per_point_display": "7.166", "released_on": "2026-06-30" }, { "rank": 24, "model_uid": "google/gemini-3.6-flash", "display": "Gemini 3.6 Flash (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.4855875831485588, "score_display": "0.486", "cost_usd_per_task": "3.532355313192905", "cost_display": "3.53236", "usd_per_point": "7.2743938184931505", "usd_per_point_display": "7.274", "released_on": "2026-07-21" }, { "rank": 25, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.6438053097345133, "score_display": "0.644", "cost_usd_per_task": "5.100437004424779", "cost_display": "5.10044", "usd_per_point": "7.922328268041237", "usd_per_point_display": "7.922", "released_on": "2026-04-23" }, { "rank": 26, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.5176991150442478, "score_display": "0.518", "cost_usd_per_task": "4.281940017699115", "cost_display": "4.28194", "usd_per_point": "8.271097811965811", "usd_per_point_display": "8.271", "released_on": "2026-05-28" }, { "rank": 27, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.7282850779510023, "score_display": "0.728", "cost_usd_per_task": "6.076084489977728", "cost_display": "6.07608", "usd_per_point": "8.34300286238532", "usd_per_point_display": "8.343", "released_on": "2026-07-24" }, { "rank": 28, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.6536697247706422, "score_display": "0.654", "cost_usd_per_task": "6.088186528935186", "cost_display": "6.08819", "usd_per_point": "9.31385728637102", "usd_per_point_display": "9.314", "released_on": "2026-06-09" }, { "rank": 29, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.3977777777777778, "score_display": "0.398", "cost_usd_per_task": "4.079036229", "cost_display": "4.07904", "usd_per_point": "10.254560352234636", "usd_per_point_display": "10.25", "released_on": "2026-06-30" }, { "rank": 30, "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.6703539823008849, "score_display": "0.67", "cost_usd_per_task": "7.226236674778761", "cost_display": "7.22624", "usd_per_point": "10.779732597359736", "usd_per_point_display": "10.78", "released_on": "2026-04-23" }, { "rank": 31, "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.5176991150442478, "score_display": "0.518", "cost_usd_per_task": "5.6524577245575225", "cost_display": "5.65246", "usd_per_point": "10.918422613247863", "usd_per_point_display": "10.92", "released_on": "2026-03-05" }, { "rank": 32, "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.7266666666666667, "score_display": "0.727", "cost_usd_per_task": "8.386436346666667", "cost_display": "8.38644", "usd_per_point": "11.540967449541284", "usd_per_point_display": "11.54", "released_on": "2026-07-09" }, { "rank": 33, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.7315436241610739, "score_display": "0.732", "cost_usd_per_task": "9.07220232326622", "cost_display": "9.0722", "usd_per_point": "12.401450882262997", "usd_per_point_display": "12.4", "released_on": "2026-07-24" }, { "rank": 34, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.6860465116279071, "score_display": "0.686", "cost_usd_per_task": "9.177638214788733", "cost_display": "9.17764", "usd_per_point": "13.377574346980184", "usd_per_point_display": "13.38", "released_on": "2026-06-09" }, { "rank": 35, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.5436241610738255, "score_display": "0.544", "cost_usd_per_task": "8.006355704138702", "cost_display": "8.00636", "usd_per_point": "14.727740739711933", "usd_per_point_display": "14.73", "released_on": "2026-05-28" }, { "rank": 36, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.4823008849557522, "score_display": "0.482", "cost_usd_per_task": "7.425560161197339", "cost_display": "7.42556", "usd_per_point": "15.396115563583475", "usd_per_point_display": "15.4", "released_on": "2026-06-30" }, { "rank": 37, "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.7364864864864865, "score_display": "0.736", "cost_usd_per_task": "11.837583271396396", "cost_display": "11.8376", "usd_per_point": "16.073048845565747", "usd_per_point_display": "16.07", "released_on": "2026-07-24" }, { "rank": 38, "model_uid": "anthropic/claude-sonnet-4-6", "display": "Claude Sonnet 4.6 (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.29933481152993346, "score_display": "0.299", "cost_usd_per_task": "5.5223517987804875", "cost_display": "5.52235", "usd_per_point": "18.448745638888887", "usd_per_point_display": "18.45", "released_on": "2026-02-17" }, { "rank": 39, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.6991150442477876, "score_display": "0.699", "cost_usd_per_task": "13.414521495535714", "cost_display": "13.4145", "usd_per_point": "19.187859860702986", "usd_per_point_display": "19.19", "released_on": "2026-06-09" }, { "rank": 40, "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (medium)", "config_key": "e4970037c54d508bd49a9dbdd947341bb7df4660ea5ca786d7566905a6818265", "effort": "medium", "score": 0.37389380530973454, "score_display": "0.374", "cost_usd_per_task": "7.341823985508849", "cost_display": "7.34182", "usd_per_point": "19.636120955325442", "usd_per_point_display": "19.64", "released_on": "2026-05-19" }, { "rank": 41, "model_uid": "anthropic/claude-opus-4-8", "display": "Claude Opus 4.8 (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.5897435897435898, "score_display": "0.59", "cost_usd_per_task": "13.222583593240094", "cost_display": "13.2226", "usd_per_point": "22.420902614624506", "usd_per_point_display": "22.42", "released_on": "2026-05-28" }, { "rank": 42, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (xhigh)", "config_key": "39fdae6337dbe72aebe28f9f8b57bfddad67ee82cd60d94cfe2f64cc58acc566", "effort": "xhigh", "score": 0.49667405764966743, "score_display": "0.497", "cost_usd_per_task": "11.890634540022173", "cost_display": "11.8906", "usd_per_point": "23.940518649776784", "usd_per_point_display": "23.94", "released_on": "2026-06-30" }, { "rank": 43, "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.6972477064220184, "score_display": "0.697", "cost_usd_per_task": "21.634702092592594", "cost_display": "21.6347", "usd_per_point": "31.028717474902535", "usd_per_point_display": "31.03", "released_on": "2026-06-09" }, { "rank": 44, "model_uid": "anthropic/claude-sonnet-5", "display": "Claude Sonnet 5 (max)", "config_key": "a869064a4fe5eec9d3bfabc7fd1bfa34e9cdb712960eb3af6be093e02db71317", "effort": "max", "score": 0.5384615384615384, "score_display": "0.538", "cost_usd_per_task": "26.399858950791852", "cost_display": "26.3999", "usd_per_point": "49.02830948004201", "usd_per_point_display": "49.03", "released_on": "2026-06-30" }, { "rank": 45, "model_uid": "google/gemini-3.1-pro-preview", "display": "Gemini 3.1 Pro Preview (high)", "config_key": "39b0bbc44fef028f30e7812c91925c018dead7387c9b588734e0f0fb2dd52eac", "effort": "high", "score": 0.11751662971175167, "score_display": "0.118", "cost_usd_per_task": "9.48151874855876", "cost_display": "9.48152", "usd_per_point": "80.68235765283019", "usd_per_point_display": "80.68", "released_on": "2026-02-19" } ], "excluded_unbuyable": 0, "excluded_no_cost": 0, "excluded_zero_cost": 0, "excluded_nonpositive_score": 0, "note": "45 ranked, 0 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score" } ] }, "f1": { "headline": { "pool_id": "baseline_2026_08_15", "rule": "the headline is a BAND, never a single number: this pool's curves run from 25–394 days depending on the threshold, so a lone median would be a choice presented as a measurement.", "curves": 152, "benchmarks": 6, "median_days": 73.731033690307, "band_days": [ 55.37947408657764, 103.36611425192395 ], "display": "median T½ 74 d · IQR 55–103 · 152 curves over 6 benchmarks", "x_axis_note": "the x axis is the MODEL'S RELEASE DATE, not the day anyone measured it or could have bought it (§5-F1 trap 6)." }, "gate": { "min_steps": 4, "min_span_days": 90, "note": "a staircase carries a half-life only with at least 4 steps AND at least 90 days between its first and its last; anything shorter is a curve whose whole life is briefer than the number it would claim" }, "corpus": { "snapshot_id": "epoch_zip@506703a3c54b4021", "fetched_at": "2026-08-17T06:11:03Z", "result_rows": 781, "points": 764, "excluded_contested": 17, "excluded_undated": 0, "release_date_disputes": 2, "cost_basis_list_price_x_tokens": 66, "cost_basis_unknown": 698, "cost_basis_note": "'imputed_local' is a price nobody can buy; WeirdML's local and quantized runs are invisible through the mirror and may sit on the frontier edge.", "zero_cost_excluded": 1, "zero_cost_note": "a published cost of $0 is not a measured spend of zero, it is a price nobody can buy — the ':free' hazard class — and one such point would pin a threshold's frontier at $0 forever. They are excluded and counted, never plotted.", "availability_filter": "OFF by decision (S29): the filter does not enter F1, F2 or the event fold, and the surface marks the stale frontier instead.", "availability_models_with_windows": 171, "measurer_note": "every measurer is its own channel: two measurers never share a frontier or a dominance timeline." }, "pools": [ { "id": "baseline_2026_08_15", "label": "the 2026-08-15 set (ALE·WeirdML·ARC×2·Proof·DeepSWE)", "note": "the same six benchmarks the 2026-08-15 ad-hoc measurement pooled, so the comparison against it is about the pipeline rather than about a different mix", "benchmarks": 6, "curves": 152, "median_days": 73.731033690307, "iqr_days": [ 55.37947408657764, 103.36611425192395 ], "p10_p90_days": [ 42.56696911738192, 159.78322062848784 ], "full_band_days": [ 25.200916416662086, 393.9802194407944 ], "per_year": { "median": 30.91908337668507, "low": 11.560547865619956, "high": 96.39521331657264 }, "display": { "median_days": "74", "iqr_days": "55–103", "p10_p90_days": "43–160", "full_band_days": "25–394", "per_year": "≈31×/yr at the median (IQR band 12×–96×)" }, "identity": { "claim": "a pool's curve count is the sum of its benchmarks' curve counts; nothing else enters a pooled median", "terms": [ { "Benchmark": "ale-bench", "Curves": 15 }, { "Benchmark": "arc-agi", "Curves": 37 }, { "Benchmark": "arc-agi-2", "Curves": 40 }, { "Benchmark": "deepswe", "Curves": 8 }, { "Benchmark": "proofbench", "Curves": 8 }, { "Benchmark": "weirdml", "Curves": 44 } ], "total": 152, "holds": true } }, { "id": "gate_no_arc_mirrors", "label": "everything past the gate, ARC mirrors excluded (§13-S20 open)", "note": "ARC-AGI and ARC-AGI-2 reach this archive only through Epoch's mirror; whether they may be pooled is S20, and it is open", "benchmarks": 6, "curves": 78, "median_days": 73.731033690307, "iqr_days": [ 57.53775247202576, 98.10887817927377 ], "p10_p90_days": [ 44.037154450048, 131.4684453860275 ], "full_band_days": [ 25.200916416662086, 393.9802194407944 ], "per_year": { "median": 30.91908337668507, "low": 13.180712673897695, "high": 81.21425452790692 }, "display": { "median_days": "74", "iqr_days": "58–98", "p10_p90_days": "44–131", "full_band_days": "25–394", "per_year": "≈31×/yr at the median (IQR band 13×–81×)" }, "identity": { "claim": "a pool's curve count is the sum of its benchmarks' curve counts; nothing else enters a pooled median", "terms": [ { "Benchmark": "ale-bench", "Curves": 15 }, { "Benchmark": "cursorbench", "Curves": 2 }, { "Benchmark": "deepswe", "Curves": 8 }, { "Benchmark": "proofbench", "Curves": 8 }, { "Benchmark": "the-agent-company", "Curves": 1 }, { "Benchmark": "weirdml", "Curves": 44 } ], "total": 78, "holds": true } }, { "id": "gate_all", "label": "everything past the gate, ARC mirrors included", "note": "the widest pool, printed beside the other two so the mirrors' effect on the band is visible rather than argued", "benchmarks": 8, "curves": 155, "median_days": 73.731033690307, "iqr_days": [ 55.37947408657764, 103.66918305660647 ], "p10_p90_days": [ 42.74169633594678, 159.76221001377343 ], "full_band_days": [ 25.200916416662086, 393.9802194407944 ], "per_year": { "median": 30.91908337668507, "low": 11.478123185543343, "high": 96.39521331657264 }, "display": { "median_days": "74", "iqr_days": "55–104", "p10_p90_days": "43–160", "full_band_days": "25–394", "per_year": "≈31×/yr at the median (IQR band 11×–96×)" }, "identity": { "claim": "a pool's curve count is the sum of its benchmarks' curve counts; nothing else enters a pooled median", "terms": [ { "Benchmark": "ale-bench", "Curves": 15 }, { "Benchmark": "arc-agi", "Curves": 37 }, { "Benchmark": "arc-agi-2", "Curves": 40 }, { "Benchmark": "cursorbench", "Curves": 2 }, { "Benchmark": "deepswe", "Curves": 8 }, { "Benchmark": "proofbench", "Curves": 8 }, { "Benchmark": "the-agent-company", "Curves": 1 }, { "Benchmark": "weirdml", "Curves": 44 } ], "total": 155, "holds": true } } ], "stale_frontier": { "decision": "S29: the availability filter stays out of the formulas; the surface MARKS a curve whose last step is held by a configuration that has left the market.", "curves": 155, "stale": 1, "unknown_holder": 0, "display": "1 of 155 valid curves (gate_all: every curve past the gate, ARC mirrors included) is held by a retired config at the corpus's now; every holder has availability data", "holders": [ { "benchmark": "weirdml", "q": 0.2318, "model_uid": "google/gemini-1.5-flash-002", "display": "Gemini 1.5 Flash 002", "at": "2026-07-31" } ] }, "benchmarks": [ { "benchmark": "ale-bench", "domain": "coding", "points": 66, "thresholds": 66, "curves": 15, "duplicate_thresholds": 33, "rejected_staircases": 18, "zero_cost_excluded": 0, "newest_release": "2026-07-31", "in_pools": [ "baseline_2026_08_15", "gate_no_arc_mirrors", "gate_all" ], "median_days": 98.10887817927377, "iqr_days": [ 76.7441761329257, 161.59776662644836 ], "full_band_days": [ 55.37947408657764, 393.9802194407944 ], "grid": [ { "q": 2176.88, "q_count": 2, "q_values": [ 2176.88, 2164.57 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "1.534078", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "availability": "on_sale" } }, { "q": 2041.3, "q_count": 1, "q_values": [ 2041.3 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "2.108822", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.534078", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "availability": "on_sale" } }, { "q": 1951.42, "q_count": 1, "q_values": [ 1951.42 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "2.108822", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.254047", "config": "GPT-5.6 Terra (max)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (max)", "availability": "on_sale" } }, { "q": 1942.97, "q_count": 1, "q_values": [ 1942.97 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "1.517132", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.254047", "config": "GPT-5.6 Terra (max)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (max)", "availability": "on_sale" } }, { "q": 1667.4, "q_count": 1, "q_values": [ 1667.4 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "1.517132", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.359171", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 1655.22, "q_count": 2, "q_values": [ 1655.22, 1607 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "0.559265", "config": "GPT-5.3-Codex (xhigh)", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.359171", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 1589.38, "q_count": 2, "q_values": [ 1589.38, 1563.83 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "0.559265", "config": "GPT-5.3-Codex (xhigh)", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.328965", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (medium)", "availability": "on_sale" } }, { "q": 1520.72, "q_count": 3, "q_values": [ 1520.72, 1463.12, 1411.83 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "0.559265", "config": "GPT-5.3-Codex (xhigh)", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.26258", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (medium)", "availability": "on_sale" } }, { "q": 1367.2, "q_count": 4, "q_values": [ 1367.2, 1327.3, 1323.05, 1309.28 ], "steps": [ { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview", "availability": "on_sale" } }, { "q": 1306.08, "q_count": 2, "q_values": [ 1306.08, 1299.9 ], "steps": [ { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1293.55, "q_count": 1, "q_values": [ 1293.55 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "0.296865", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1249.83, "q_count": 1, "q_values": [ 1249.83 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1244.92, "q_count": 1, "q_values": [ 1244.92 ], "steps": [ { "released_on": "2025-11-12", "cost_usd_per_task": "0.512093", "config": "GPT-5.1-Codex", "model_uid": "openai/gpt-5.1-codex" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 261, "ratio": 26.226211205572056, "t_half_days": 55.37947408657764, "t_half_ols_days": 68.86638944451552, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 14.048294807385801, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1208.83, "q_count": 1, "q_values": [ 1208.83 ], "steps": [ { "released_on": "2025-11-12", "cost_usd_per_task": "0.512093", "config": "GPT-5.1-Codex", "model_uid": "openai/gpt-5.1-codex" }, { "released_on": "2025-11-19", "cost_usd_per_task": "0.253186", "config": "GPT-5.1-Codex-Max", "model_uid": "openai/gpt-5.1-codex-max" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 261, "ratio": 26.226211205572056, "t_half_days": 55.37947408657764, "t_half_ols_days": 67.90418375659293, "t_half_drop_first_days": 68.70938604722686, "t_half_drop_last_days": 14.048294807385801, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1192.15, "q_count": 2, "q_values": [ 1192.15, 1188.58 ], "steps": [ { "released_on": "2025-11-12", "cost_usd_per_task": "0.512093", "config": "GPT-5.1-Codex", "model_uid": "openai/gpt-5.1-codex" }, { "released_on": "2025-11-13", "cost_usd_per_task": "0.315236", "config": "GPT-5.1 (high)", "model_uid": "openai/gpt-5.1-20251113" }, { "released_on": "2025-11-19", "cost_usd_per_task": "0.253186", "config": "GPT-5.1-Codex-Max", "model_uid": "openai/gpt-5.1-codex-max" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 261, "ratio": 26.226211205572056, "t_half_days": 55.37947408657764, "t_half_ols_days": 66.24995203436292, "t_half_drop_first_days": 64.79001622182983, "t_half_drop_last_days": 14.048294807385801, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1176.75, "q_count": 1, "q_values": [ 1176.75 ], "steps": [ { "released_on": "2025-11-12", "cost_usd_per_task": "0.512093", "config": "GPT-5.1-Codex", "model_uid": "openai/gpt-5.1-codex" }, { "released_on": "2025-11-13", "cost_usd_per_task": "0.315236", "config": "GPT-5.1 (high)", "model_uid": "openai/gpt-5.1-20251113" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.259385", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-11-19", "cost_usd_per_task": "0.253186", "config": "GPT-5.1-Codex-Max", "model_uid": "openai/gpt-5.1-codex-max" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 261, "ratio": 26.226211205572056, "t_half_days": 55.37947408657764, "t_half_ols_days": 66.02563350634559, "t_half_drop_first_days": 64.79001622182983, "t_half_drop_last_days": 14.048294807385801, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1162.45, "q_count": 3, "q_values": [ 1162.45, 1160.6, 1150.28 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.244948", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 12.544709617945303, "t_half_days": 98.10887817927377, "t_half_ols_days": 96.34898921238782, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 92.47095537677505, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1127.58, "q_count": 1, "q_values": [ 1127.58 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.244948", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.073484", "config": "GPT-5.5 (none)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 12.544709617945303, "t_half_days": 98.10887817927377, "t_half_ols_days": 105.59773230979388, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 149.10993625675, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1086.03, "q_count": 3, "q_values": [ 1086.03, 1025.38, 1006.08 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.244948", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.052314", "config": "GPT-5.4 (none)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 12.544709617945303, "t_half_days": 98.10887817927377, "t_half_ols_days": 95.26242855475708, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 94.28851698253911, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 1004.52, "q_count": 1, "q_values": [ 1004.52 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.244948", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.052314", "config": "GPT-5.4 (none)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 12.544709617945303, "t_half_days": 98.10887817927377, "t_half_ols_days": 89.8212023987349, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 69.87806127807583, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 996.5, "q_count": 2, "q_values": [ 996.5, 944.17 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.244948", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.142947", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-02-05", "cost_usd_per_task": "0.061688", "config": "Claude Opus 4.6", "model_uid": "anthropic/claude-opus-4-6-20260205" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.052314", "config": "GPT-5.4 (none)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 12.544709617945303, "t_half_days": 98.10887817927377, "t_half_ols_days": 89.79897608926646, "t_half_drop_first_days": 80.77960672778048, "t_half_drop_last_days": 69.87806127807583, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 933.55, "q_count": 2, "q_values": [ 933.55, 911.02 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.103474", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.091067", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-02-05", "cost_usd_per_task": "0.061688", "config": "Claude Opus 4.6", "model_uid": "anthropic/claude-opus-4-6-20260205" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.052314", "config": "GPT-5.4 (none)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 5.299293250025606, "t_half_days": 195.77687541985108, "t_half_ols_days": 194.08647915108506, "t_half_drop_first_days": 101.7316013875296, "t_half_drop_last_days": 173.23802659353981, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 826.17, "q_count": 1, "q_values": [ 826.17 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 807.65, "q_count": 1, "q_values": [ 807.65 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.043744", "config": "GPT-5 (minimal)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 2.290228413397521, "t_half_days": 393.9802194407944, "t_half_ols_days": 376.0501593119339, "t_half_drop_first_days": 307.6424101423852, "t_half_drop_last_days": 463.0608080225178, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 804.12, "q_count": 1, "q_values": [ 804.12 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.043744", "config": "GPT-5 (minimal)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.027084", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 2.290228413397521, "t_half_days": 393.9802194407944, "t_half_ols_days": 381.5382181472995, "t_half_drop_first_days": 368.13391923631275, "t_half_drop_last_days": 463.0608080225178, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 799.77, "q_count": 6, "q_values": [ 799.77, 797.73, 796.15, 785.52, 765.27, 763.98 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.020554", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.019526", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 2.290228413397521, "t_half_days": 393.9802194407944, "t_half_ols_days": 427.5117728548451, "t_half_drop_first_days": 368.13391923631275, "t_half_drop_last_days": 100.76072168579894, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 745.17, "q_count": 1, "q_values": [ 745.17 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.020554", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-22", "cost_usd_per_task": "0.01883", "config": "DeepSeek-V3.1-Terminus", "model_uid": "deepseek/deepseek-v3.1-terminus" } ], "valid": true, "span_days": 159, "ratio": 2.374880509824748, "t_half_days": 127.41865783304564, "t_half_ols_days": 109.85759339089519, "t_half_drop_first_days": 96.08276110768229, "t_half_drop_last_days": 100.76072168579894, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v3.1-terminus", "display": "DeepSeek-V3.1-Terminus", "availability": "on_sale" } }, { "q": 718.67, "q_count": 2, "q_values": [ 718.67, 715.52 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.007194", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 678.2, "q_count": 6, "q_values": [ 678.2, 674.77, 661.88, 655.35, 653.48, 587.73 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.007194", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.005179", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 373, "ratio": 8.634678509364743, "t_half_days": 119.93019752744037, "t_half_ols_days": 120.51163632848473, "t_half_drop_first_days": 107.46795943459871, "t_half_drop_last_days": 42.867626252550714, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 575.62, "q_count": 1, "q_values": [ 575.62 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.001303", "config": "gpt-oss-120b", "model_uid": "openai/gpt-oss-120b" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-120b", "display": "gpt-oss-120b", "availability": "on_sale" } }, { "q": 566.05, "q_count": 1, "q_values": [ 566.05 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.044719", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.043794", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.000717", "config": "gpt-oss-20b", "model_uid": "openai/gpt-oss-20b" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b", "availability": "on_sale" } }, { "q": 558.1, "q_count": 7, "q_values": [ 558.1, 521.67, 497.62, 394.93, 324.98, 264.7, 210.18 ], "steps": [ { "released_on": "2025-04-14", "cost_usd_per_task": "0.026917", "config": "GPT-4.1", "model_uid": "openai/gpt-4.1-20250414" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.000717", "config": "gpt-oss-20b", "model_uid": "openai/gpt-oss-20b" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b", "availability": "on_sale" } }, { "q": 137.78, "q_count": 1, "q_values": [ 137.78 ], "steps": [ { "released_on": "2025-04-14", "cost_usd_per_task": "0.026917", "config": "GPT-4.1", "model_uid": "openai/gpt-4.1-20250414" }, { "released_on": "2025-07-30", "cost_usd_per_task": "0.00278", "config": "Codestral (2508)", "model_uid": "mistral/codestral-2508" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.000717", "config": "gpt-oss-20b", "model_uid": "openai/gpt-oss-20b" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b", "availability": "on_sale" } } ] }, { "benchmark": "arc-agi", "domain": "reasoning", "points": 138, "thresholds": 118, "curves": 37, "duplicate_thresholds": 62, "rejected_staircases": 19, "zero_cost_excluded": 0, "newest_release": "2026-07-09", "in_pools": [ "baseline_2026_08_15", "gate_all" ], "median_days": 72.05662362803477, "iqr_days": [ 52.88487685740664, 96.39725966688523 ], "full_band_days": [ 35.36123418715585, 230.9678582683722 ], "grid": [ { "q": 0.98, "q_count": 1, "q_values": [ 0.98 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.1-pro-preview", "display": "Gemini 3.1 Pro Preview", "availability": "on_sale" } }, { "q": 0.975, "q_count": 1, "q_values": [ 0.975 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.4", "config": "GPT-5.6 Sol (xhigh)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (xhigh)", "availability": "on_sale" } }, { "q": 0.97, "q_count": 4, "q_values": [ 0.97, 0.965, 0.9500000000000001, 0.9450000000000001 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.9400000000000001, "q_count": 3, "q_values": [ 0.9400000000000001, 0.9367000000000001, 0.935 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.26", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.9267000000000001, "q_count": 1, "q_values": [ 0.9267000000000001 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.37", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.26", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.925, "q_count": 2, "q_values": [ 0.925, 0.9217000000000001 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.37", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.22", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.92, "q_count": 3, "q_values": [ 0.92, 0.915, 0.91 ], "steps": [ { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.37", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.905, "q_count": 1, "q_values": [ 0.905 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "11.6542", "config": "GPT-5.2 pro (xhigh)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.5223", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.37", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 61.3378947368421, "t_half_days": 35.36123418715585, "t_half_ols_days": 39.68919752848033, "t_half_drop_first_days": 95.96407422100492, "t_half_drop_last_days": 16.877023720625296, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.895, "q_count": 3, "q_values": [ 0.895, 0.88, 0.8767 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "11.6542", "config": "GPT-5.2 pro (xhigh)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.35", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.8717, "q_count": 2, "q_values": [ 0.8717, 0.865 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "11.6542", "config": "GPT-5.2 pro (xhigh)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.35", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-07-08", "cost_usd_per_task": "0.31", "config": "Grok 4.5 (medium)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 61.3378947368421, "t_half_days": 35.36123418715585, "t_half_ols_days": 46.71739487842539, "t_half_drop_first_days": 161.11546298386222, "t_half_drop_last_days": 39.94313970864375, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.8617, "q_count": 2, "q_values": [ 0.8617, 0.86 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "11.6542", "config": "GPT-5.2 pro (xhigh)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.35", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.25", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 61.3378947368421, "t_half_days": 35.36123418715585, "t_half_ols_days": 41.115856686131124, "t_half_drop_first_days": 161.11546298386222, "t_half_drop_last_days": 15.154855372205429, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.8567, "q_count": 1, "q_values": [ 0.8567 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "5.8694", "config": "GPT-5.2 pro (high)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.35", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.25", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 30.89157894736842, "t_half_days": 42.43159974167465, "t_half_ols_days": 48.94521251605932, "t_half_drop_first_days": 161.11546298386222, "t_half_drop_last_days": 18.448510848027574, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.8467, "q_count": 1, "q_values": [ 0.8467 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "5.8694", "config": "GPT-5.2 pro (high)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (high)", "availability": "on_sale" } }, { "q": 0.8117000000000001, "q_count": 1, "q_values": [ 0.8117000000000001 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "3.9774", "config": "GPT-5.2 pro (medium)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (high)", "availability": "on_sale" } }, { "q": 0.8, "q_count": 2, "q_values": [ 0.8, 0.7917000000000001 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.47", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (high)", "availability": "on_sale" } }, { "q": 0.787, "q_count": 1, "q_values": [ 0.787 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.47", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.5196", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (high)", "availability": "on_sale" } }, { "q": 0.77, "q_count": 3, "q_values": [ 0.77, 0.765, 0.7617 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.47", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.5196", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 227, "ratio": 11.307692307692307, "t_half_days": 64.87136587268913, "t_half_ols_days": 98.21352160800102, "t_half_drop_first_days": 105.05831747637815, "t_half_drop_last_days": 7.476825869605145, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.75, "q_count": 2, "q_values": [ 0.75, 0.745 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.7267, "q_count": 2, "q_values": [ 0.7267, 0.72 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 233, "ratio": 3.789230769230769, "t_half_days": 121.23387963732218, "t_half_ols_days": 158.7390999359081, "t_half_drop_first_days": 149.2274122548424, "t_half_drop_last_days": 19.34824362680815, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.7017, "q_count": 1, "q_values": [ 0.7017 ], "steps": [ { "released_on": "2025-10-07", "cost_usd_per_task": "4.7785", "config": "GPT-5 pro (unknown)", "model_uid": "openai/gpt-5-pro-20251006" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 275, "ratio": 36.75769230769231, "t_half_days": 52.88487685740664, "t_half_ols_days": 74.91822958216015, "t_half_drop_first_days": 121.23387963732218, "t_half_drop_last_days": 14.863154277022536, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.6817000000000001, "q_count": 1, "q_values": [ 0.6817000000000001 ], "steps": [ { "released_on": "2025-10-07", "cost_usd_per_task": "4.7785", "config": "GPT-5 pro (unknown)", "model_uid": "openai/gpt-5-pro-20251006" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 275, "ratio": 36.75769230769231, "t_half_days": 52.88487685740664, "t_half_ols_days": 70.64209765296188, "t_half_drop_first_days": 121.23387963732218, "t_half_drop_last_days": 29.838650524689793, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.6667000000000001, "q_count": 1, "q_values": [ 0.6667000000000001 ], "steps": [ { "released_on": "2025-07-09", "cost_usd_per_task": "2.17", "config": "Grok 4 (0709)", "model_uid": "xai/grok-4-0709" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 365, "ratio": 16.69230769230769, "t_half_days": 89.87687206186196, "t_half_ols_days": 89.69090951618182, "t_half_drop_first_days": 121.23387963732218, "t_half_drop_last_days": 62.00286419926696, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.657, "q_count": 2, "q_values": [ 0.657, 0.6367 ], "steps": [ { "released_on": "2025-07-09", "cost_usd_per_task": "2.17", "config": "Grok 4 (0709)", "model_uid": "xai/grok-4-0709" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.5087", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 365, "ratio": 16.69230769230769, "t_half_days": 89.87687206186196, "t_half_ols_days": 102.2433478360449, "t_half_drop_first_days": 170.7053840830622, "t_half_drop_last_days": 62.00286419926696, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.608, "q_count": 1, "q_values": [ 0.608 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.5002", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.13", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 449, "ratio": 3.8476923076923075, "t_half_days": 230.9678582683722, "t_half_ols_days": 206.3402907002912, "t_half_drop_first_days": 121.23387963732218, "t_half_drop_last_days": 185.89472743348844, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.6017, "q_count": 2, "q_values": [ 0.6017, 0.593 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.5002", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.4926", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.09", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 449, "ratio": 5.557777777777778, "t_half_days": 181.450200446943, "t_half_ols_days": 171.19565739774066, "t_half_drop_first_days": 95.0082067837094, "t_half_drop_last_days": 185.89472743348844, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.587, "q_count": 1, "q_values": [ 0.587 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.09", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 449, "ratio": 4.508888888888889, "t_half_days": 206.6484695898712, "t_half_ols_days": 205.8973146098565, "t_half_drop_first_days": 108.37239348621114, "t_half_drop_last_days": 224.96069666118024, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.5867, "q_count": 2, "q_values": [ 0.5867, 0.58 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.38", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.1743", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.15", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.09", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 449, "ratio": 4.508888888888889, "t_half_days": 206.6484695898712, "t_half_ols_days": 198.7608685308003, "t_half_drop_first_days": 109.23952145822089, "t_half_drop_last_days": 224.96069666118024, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.5767, "q_count": 1, "q_values": [ 0.5767 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.38", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.3448", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 245, "ratio": 6.630718954248366, "t_half_days": 89.77103708810043, "t_half_ols_days": 203.06591954433736, "t_half_drop_first_days": 8.730654471727972, "t_half_drop_last_days": 1016.9819359800193, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (medium)", "availability": "on_sale" } }, { "q": 0.57, "q_count": 1, "q_values": [ 0.57 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.38", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.08", "config": "DeepSeek-V3.2", "model_uid": "deepseek/deepseek-v3.2" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 245, "ratio": 6.630718954248366, "t_half_days": 89.77103708810043, "t_half_ols_days": 124.2358654361665, "t_half_drop_first_days": 8.730654471727972, "t_half_drop_last_days": 97.75058571884855, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (medium)", "availability": "on_sale" } }, { "q": 0.565, "q_count": 1, "q_values": [ 0.565 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.38", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.08", "config": "DeepSeek-V3.2", "model_uid": "deepseek/deepseek-v3.2" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.06", "config": "GPT-5.6 Luna (medium)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 449, "ratio": 6.763333333333334, "t_half_days": 162.81480564316414, "t_half_ols_days": 155.67992386712416, "t_half_drop_first_days": 85.24332798797496, "t_half_drop_last_days": 89.77103708810043, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "availability": "on_sale" } }, { "q": 0.562, "q_count": 2, "q_values": [ 0.562, 0.5567 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.3301", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.08", "config": "DeepSeek-V3.2", "model_uid": "deepseek/deepseek-v3.2" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.06", "config": "GPT-5.6 Luna (medium)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 449, "ratio": 6.763333333333334, "t_half_days": 162.81480564316414, "t_half_ols_days": 141.4597483140092, "t_half_drop_first_days": 136.59265453061755, "t_half_drop_last_days": 89.77103708810043, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "availability": "on_sale" } }, { "q": 0.5433, "q_count": 1, "q_values": [ 0.5433 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.4058", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.116", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.08", "config": "DeepSeek-V3.2", "model_uid": "deepseek/deepseek-v3.2" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.06", "config": "GPT-5.6 Luna (medium)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 449, "ratio": 6.763333333333334, "t_half_days": 162.81480564316414, "t_half_ols_days": 172.34965479563675, "t_half_drop_first_days": 353.27872110780226, "t_half_drop_last_days": 89.77103708810043, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "availability": "on_sale" } }, { "q": 0.538, "q_count": 3, "q_values": [ 0.538, 0.515, 0.4883 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.2882", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.116", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.08", "config": "DeepSeek-V3.2", "model_uid": "deepseek/deepseek-v3.2" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0612", "config": "Gemini 3 Flash Preview (medium)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.06", "config": "GPT-5.6 Luna (medium)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 449, "ratio": 4.803333333333334, "t_half_days": 198.31840745729093, "t_half_ols_days": 204.48828726093274, "t_half_drop_first_days": 353.27872110780226, "t_half_drop_last_days": 109.59679762092203, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "availability": "on_sale" } }, { "q": 0.485, "q_count": 6, "q_values": [ 0.485, 0.4833, 0.4767, 0.465, 0.443, 0.44 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.2882", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.116", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.418, "q_count": 6, "q_values": [ 0.418, 0.415, 0.41000000000000003, 0.4083, 0.4, 0.38170000000000004 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.116", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.3733000000000001, "q_count": 3, "q_values": [ 0.3733000000000001, 0.37, 0.357 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0401", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.345, "q_count": 2, "q_values": [ 0.345, 0.3417 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.3989", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0401", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 231, "ratio": 12.993485342019543, "t_half_days": 62.43721541478278, "t_half_ols_days": 61.35801200321488, "t_half_drop_first_days": 68.16239606932523, "t_half_drop_last_days": 56.722986670965355, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.333, "q_count": 7, "q_values": [ 0.333, 0.33170000000000005, 0.33, 0.3233, 0.31, 0.307, 0.295 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.3989", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 231, "ratio": 12.993485342019543, "t_half_days": 62.43721541478278, "t_half_ols_days": 62.04846251409361, "t_half_drop_first_days": 68.16239606932523, "t_half_drop_last_days": 31.810551238638013, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.29, "q_count": 1, "q_values": [ 0.29 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.3989", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0163", "config": "Gemini 3 Flash Preview (low)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 320, "ratio": 24.472392638036812, "t_half_days": 69.36792214458981, "t_half_ols_days": 74.36650968016659, "t_half_drop_first_days": 76.51423312044969, "t_half_drop_last_days": 62.43721541478278, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (low)", "availability": "on_sale" } }, { "q": 0.286, "q_count": 3, "q_values": [ 0.286, 0.28, 0.27 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.3989", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-02-24", "cost_usd_per_task": "0.33", "config": "Claude 3.7 Sonnet", "model_uid": "anthropic/claude-3-7-sonnet-20250219" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.0307", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.0163", "config": "Gemini 3 Flash Preview (low)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 320, "ratio": 24.472392638036812, "t_half_days": 69.36792214458981, "t_half_ols_days": 69.57897979528123, "t_half_drop_first_days": 68.21027510266916, "t_half_drop_last_days": 62.43721541478278, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (low)", "availability": "on_sale" } }, { "q": 0.2633, "q_count": 5, "q_values": [ 0.2633, 0.258, 0.255, 0.238, 0.225 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.3989", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-02-24", "cost_usd_per_task": "0.33", "config": "Claude 3.7 Sonnet", "model_uid": "anthropic/claude-3-7-sonnet-20250219" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0135", "config": "GPT-5 mini (low)", "model_uid": "openai/gpt-5-mini-20250807" } ], "valid": true, "span_days": 188, "ratio": 29.548148148148147, "t_half_days": 38.48519165057719, "t_half_ols_days": 36.01670685519548, "t_half_drop_first_days": 35.56376924741148, "t_half_drop_last_days": 31.810551238638013, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (low)", "availability": "on_sale" } }, { "q": 0.223, "q_count": 1, "q_values": [ 0.223 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0135", "config": "GPT-5 mini (low)", "model_uid": "openai/gpt-5-mini-20250807" } ], "valid": true, "span_days": 188, "ratio": 14.125925925925927, "t_half_days": 49.2111359203903, "t_half_ols_days": 46.27347641567752, "t_half_drop_first_days": 32.527990300879644, "t_half_drop_last_days": 46.15097186300771, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (low)", "availability": "on_sale" } }, { "q": 0.215, "q_count": 1, "q_values": [ 0.215 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.15", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0135", "config": "GPT-5 mini (low)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.012", "config": "Gemini 3 Flash Preview (minimal)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 320, "ratio": 15.891666666666667, "t_half_days": 80.19651086963073, "t_half_ols_days": 72.29423020320841, "t_half_drop_first_days": 67.23646248375864, "t_half_drop_last_days": 49.2111359203903, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (minimal)", "availability": "on_sale" } }, { "q": 0.213, "q_count": 2, "q_values": [ 0.213, 0.212 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.0406", "config": "o4-mini (low)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0135", "config": "GPT-5 mini (low)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.012", "config": "Gemini 3 Flash Preview (minimal)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 320, "ratio": 15.891666666666667, "t_half_days": 80.19651086963073, "t_half_ols_days": 84.58376353121517, "t_half_drop_first_days": 139.32761911546277, "t_half_drop_last_days": 49.2111359203903, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (minimal)", "availability": "on_sale" } }, { "q": 0.2071, "q_count": 1, "q_values": [ 0.2071 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.0406", "config": "o4-mini (low)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0124", "config": "GPT-5 nano (medium)", "model_uid": "openai/gpt-5-nano-20250807" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.012", "config": "Gemini 3 Flash Preview (minimal)", "model_uid": "google/gemini-3-flash-preview" } ], "valid": true, "span_days": 320, "ratio": 15.891666666666667, "t_half_days": 80.19651086963073, "t_half_ols_days": 83.85845330660216, "t_half_drop_first_days": 139.32761911546277, "t_half_drop_last_days": 47.68072800388612, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3-flash-preview", "display": "Gemini 3 Flash Preview (minimal)", "availability": "on_sale" } }, { "q": 0.1833, "q_count": 3, "q_values": [ 0.1833, 0.16830000000000003, 0.1667 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.0406", "config": "o4-mini (low)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0371", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0124", "config": "GPT-5 nano (medium)", "model_uid": "openai/gpt-5-nano-20250807" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.012", "config": "Gemini 3 Flash Preview (minimal)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.01", "config": "GPT-5.4 nano (low)", "model_uid": "openai/gpt-5.4-nano-20260317" } ], "valid": true, "span_days": 410, "ratio": 19.07, "t_half_days": 96.39725966688523, "t_half_ols_days": 109.46974659173559, "t_half_drop_first_days": 165.72018777014424, "t_half_drop_last_days": 80.19651086963073, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-nano-20260317", "display": "GPT-5.4 nano (low)", "availability": "on_sale" } }, { "q": 0.165, "q_count": 2, "q_values": [ 0.165, 0.16 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.1907", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-3-mini", "display": "Grok 3 mini (low)", "availability": "on_sale" } }, { "q": 0.158, "q_count": 1, "q_values": [ 0.158 ], "steps": [ { "released_on": "2025-01-20", "cost_usd_per_task": "0.06", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-3-mini", "display": "Grok 3 mini (low)", "availability": "on_sale" } }, { "q": 0.145, "q_count": 2, "q_values": [ 0.145, 0.1433 ], "steps": [ { "released_on": "2025-01-20", "cost_usd_per_task": "0.06", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.0519", "config": "o3-mini (low)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "xai/grok-3-mini", "display": "Grok 3 mini (low)", "availability": "on_sale" } }, { "q": 0.14, "q_count": 9, "q_values": [ 0.14, 0.136, 0.13, 0.1233, 0.116, 0.103, 0.06, 0.058300000000000005, 0.055 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.06", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.0519", "config": "o3-mini (low)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" } ], "valid": true, "span_days": 209, "ratio": 13.636363636363637, "t_half_days": 55.44668032645097, "t_half_ols_days": 60.78105948180868, "t_half_drop_first_days": 30.390903140721395, "t_half_drop_last_days": 102.23666534708433, "stale_frontier": false, "holder": { "model_uid": "xai/grok-3-mini", "display": "Grok 3 mini (low)", "availability": "on_sale" } }, { "q": 0.0533, "q_count": 2, "q_values": [ 0.0533, 0.05 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.06", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.0519", "config": "o3-mini (low)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0057", "config": "GPT-5 mini (minimal)", "model_uid": "openai/gpt-5-mini-20250807" } ], "valid": true, "span_days": 329, "ratio": 23.68421052631579, "t_half_days": 72.05662362803477, "t_half_ols_days": 66.03202595776149, "t_half_drop_first_days": 58.59958172472389, "t_half_drop_last_days": 55.44668032645097, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (minimal)", "availability": "on_sale" } }, { "q": 0.045, "q_count": 1, "q_values": [ 0.045 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-11-20", "cost_usd_per_task": "0.05", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0057", "config": "GPT-5 mini (minimal)", "model_uid": "openai/gpt-5-mini-20250807" } ], "valid": true, "span_days": 329, "ratio": 23.68421052631579, "t_half_days": 72.05662362803477, "t_half_ols_days": 71.22945124257359, "t_half_drop_first_days": 82.99035254666451, "t_half_drop_last_days": 55.44668032645097, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (minimal)", "availability": "on_sale" } }, { "q": 0.0404, "q_count": 1, "q_values": [ 0.0404 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-11-20", "cost_usd_per_task": "0.05", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0033", "config": "GPT-5 nano (low)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 329, "ratio": 40.909090909090914, "t_half_days": 61.445371756234984, "t_half_ols_days": 61.714688005232475, "t_half_drop_first_days": 66.30301730760529, "t_half_drop_last_days": 55.44668032645097, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (low)", "availability": "on_sale" } }, { "q": 0.035, "q_count": 1, "q_values": [ 0.035 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-11-20", "cost_usd_per_task": "0.05", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.0078", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0033", "config": "GPT-5 nano (low)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 329, "ratio": 40.909090909090914, "t_half_days": 61.445371756234984, "t_half_ols_days": 60.4369762230804, "t_half_drop_first_days": 66.30301730760529, "t_half_drop_last_days": 52.025828976452914, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (low)", "availability": "on_sale" } }, { "q": 0.015, "q_count": 1, "q_values": [ 0.015 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-11-20", "cost_usd_per_task": "0.05", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.0078", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0015", "config": "GPT-5 nano (minimal)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 329, "ratio": 90, "t_half_days": 50.678904022952715, "t_half_ols_days": 51.76962502648621, "t_half_drop_first_days": 51.39463605700745, "t_half_drop_last_days": 52.025828976452914, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (minimal)", "availability": "on_sale" } }, { "q": 0, "q_count": 1, "q_values": [ 0 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.135", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-11-20", "cost_usd_per_task": "0.05", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-04-09", "cost_usd_per_task": "0.0099", "config": "Grok 3 mini (low)", "model_uid": "xai/grok-3-mini" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.0021", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0015", "config": "GPT-5 nano (minimal)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 329, "ratio": 90, "t_half_days": 50.678904022952715, "t_half_ols_days": 48.28702523208063, "t_half_drop_first_days": 51.39463605700745, "t_half_drop_last_days": 35.62850693744521, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (minimal)", "availability": "on_sale" } } ] }, { "benchmark": "arc-agi-2", "domain": "reasoning", "points": 133, "thresholds": 96, "curves": 40, "duplicate_thresholds": 42, "rejected_staircases": 14, "zero_cost_excluded": 0, "newest_release": "2026-07-09", "in_pools": [ "baseline_2026_08_15", "gate_all" ], "median_days": 86.8868445338903, "iqr_days": [ 50.992368378120695, 117.31759231144274 ], "full_band_days": [ 32.04311576027354, 223.44352227218513 ], "grid": [ { "q": 0.925, "q_count": 1, "q_values": [ 0.925 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "1.44", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "availability": "on_sale" } }, { "q": 0.9, "q_count": 1, "q_values": [ 0.9 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "1.04", "config": "GPT-5.6 Sol (xhigh)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (xhigh)", "availability": "on_sale" } }, { "q": 0.8542000000000001, "q_count": 1, "q_values": [ 0.8542000000000001 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "0.74", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.85, "q_count": 4, "q_values": [ 0.85, 0.8458, 0.8416, 0.839 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "1.87", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.74", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.8333, "q_count": 1, "q_values": [ 0.8333 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "16.41", "config": "GPT-5.4 pro (xhigh)", "model_uid": "openai/gpt-5.4-pro-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.45", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.74", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.7417, "q_count": 1, "q_values": [ 0.7417 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "16.41", "config": "GPT-5.4 pro (xhigh)", "model_uid": "openai/gpt-5.4-pro-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.45", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.69", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.7395, "q_count": 1, "q_values": [ 0.7395 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "1.52", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.45", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.69", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.7208, "q_count": 2, "q_values": [ 0.7208, 0.7167 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "1.52", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.45", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.8503", "config": "Gemini 3.5 Flash (high)", "model_uid": "google/gemini-3.5-flash" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.69", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 2.2028985507246377, "t_half_days": 110.58422150345315, "t_half_ols_days": 101.05093350513458, "t_half_drop_first_days": 71.86961396615791, "t_half_drop_last_days": 89.49587183140567, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.7042, "q_count": 1, "q_values": [ 0.7042 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "1.52", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.86", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.8503", "config": "Gemini 3.5 Flash (high)", "model_uid": "google/gemini-3.5-flash" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.69", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 2.2028985507246377, "t_half_days": 110.58422150345315, "t_half_ols_days": 115.38302240042843, "t_half_drop_first_days": 242.3362743193336, "t_half_drop_last_days": 89.49587183140567, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.675, "q_count": 1, "q_values": [ 0.675 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "1.02", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.86", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.8503", "config": "Gemini 3.5 Flash (high)", "model_uid": "google/gemini-3.5-flash" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.69", "config": "GPT-5.6 Terra (xhigh)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 1.4782608695652175, "t_half_days": 223.44352227218513, "t_half_ols_days": 231.4202254291, "t_half_drop_first_days": 242.3362743193336, "t_half_drop_last_days": 285.6867409588003, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (xhigh)", "availability": "on_sale" } }, { "q": 0.6708, "q_count": 1, "q_values": [ 0.6708 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "1.02", "config": "GPT-5.4 (high)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.86", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.8503", "config": "Gemini 3.5 Flash (high)", "model_uid": "google/gemini-3.5-flash" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 126, "ratio": 2.170212765957447, "t_half_days": 112.71773742765987, "t_half_ols_days": 116.67439398009435, "t_half_drop_first_days": 88.33558405461538, "t_half_drop_last_days": 285.6867409588003, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.6514, "q_count": 6, "q_values": [ 0.6514, 0.6222, 0.6208, 0.6042, 0.5954, 0.5833 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.86", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.8503", "config": "Gemini 3.5 Flash (high)", "model_uid": "google/gemini-3.5-flash" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 142, "ratio": 1.9574468085106385, "t_half_days": 146.5468952215661, "t_half_ols_days": 157.76834782712373, "t_half_drop_first_days": 88.33558405461538, "t_half_drop_last_days": 800.6199207465313, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.5542, "q_count": 1, "q_values": [ 0.5542 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.5416, "q_count": 1, "q_values": [ 0.5416 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "15.721", "config": "GPT-5.2 pro (high)", "model_uid": "openai/gpt-5.2-pro-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 210, "ratio": 33.44893617021277, "t_half_days": 41.47010804606171, "t_half_ols_days": 47.60442572440302, "t_half_drop_first_days": 146.5468952215661, "t_half_drop_last_days": 18.538894821421504, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.5291, "q_count": 3, "q_values": [ 0.5291, 0.5264, 0.4764 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "1.8982", "config": "GPT-5.2 (xhigh)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 210, "ratio": 4.03872340425532, "t_half_days": 104.27532066597152, "t_half_ols_days": 110.92239536470257, "t_half_drop_first_days": 146.5468952215661, "t_half_drop_last_days": 56.71746245675061, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.4333, "q_count": 1, "q_values": [ 0.4333 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.47", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 210, "ratio": 2.963829787234043, "t_half_days": 133.9744887891047, "t_half_ols_days": 137.73857363743477, "t_half_drop_first_days": 146.5468952215661, "t_half_drop_last_days": 81.19169250000759, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.425, "q_count": 2, "q_values": [ 0.425, 0.3847 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.32", "config": "GPT-5.6 Sol (low)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 210, "ratio": 4.353125, "t_half_days": 98.9608429268929, "t_half_ols_days": 98.62708370304962, "t_half_drop_first_days": 93.20264229194645, "t_half_drop_last_days": 81.19169250000759, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (low)", "availability": "on_sale" } }, { "q": 0.3764, "q_count": 1, "q_values": [ 0.3764 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "2.4", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.32", "config": "GPT-5.6 Sol (low)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 227, "ratio": 7.5, "t_half_days": 78.09031421510399, "t_half_ols_days": 84.54355901392863, "t_half_drop_first_days": 98.9608429268929, "t_half_drop_last_days": 55.511959602720665, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (low)", "availability": "on_sale" } }, { "q": 0.375, "q_count": 1, "q_values": [ 0.375 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "2.4", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.27", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 227, "ratio": 8.888888888888888, "t_half_days": 72.01769581764447, "t_half_ols_days": 77.52291458771295, "t_half_drop_first_days": 88.71375402398247, "t_half_drop_last_days": 55.511959602720665, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.3333, "q_count": 3, "q_values": [ 0.3333, 0.3306, 0.2931 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "2.4", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.68", "config": "GPT-5.4 (medium)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "0.35", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.27", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 227, "ratio": 8.888888888888888, "t_half_days": 72.01769581764447, "t_half_ols_days": 73.08383386924508, "t_half_drop_first_days": 88.71375402398247, "t_half_drop_last_days": 54.00330887352079, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.2917, "q_count": 1, "q_values": [ 0.2917 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "2.4", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "1.393", "config": "GPT-5.2 (high)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.92", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" } ], "valid": true, "span_days": 101, "ratio": 8.888888888888888, "t_half_days": 32.04311576027354, "t_half_ols_days": 41.3476012869171, "t_half_drop_first_days": 35.485501609592994, "t_half_drop_last_days": 61.44599162463533, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (low)", "availability": "on_sale" } }, { "q": 0.2667, "q_count": 1, "q_values": [ 0.2667 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "2.4", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.7594", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (low)", "availability": "on_sale" } }, { "q": 0.2278, "q_count": 2, "q_values": [ 0.2278, 0.189 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "0.79", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.7594", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-20260305", "display": "GPT-5.4 (low)", "availability": "on_sale" } }, { "q": 0.1875, "q_count": 1, "q_values": [ 0.1875 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "0.79", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.7594", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 227, "ratio": 4.647058823529411, "t_half_days": 102.42212964599804, "t_half_ols_days": 97.5158608512984, "t_half_drop_first_days": 97.25260228755457, "t_half_drop_last_days": 65.20785099937801, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.1833, "q_count": 1, "q_values": [ 0.1833 ], "steps": [ { "released_on": "2025-10-07", "cost_usd_per_task": "7.1432", "config": "GPT-5 pro (unknown)", "model_uid": "openai/gpt-5-pro-20251006" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.79", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.7594", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 275, "ratio": 42.01882352941176, "t_half_days": 50.992368378120695, "t_half_ols_days": 60.26147794106956, "t_half_drop_first_days": 102.42212964599804, "t_half_drop_last_days": 31.530793333491317, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.1764, "q_count": 1, "q_values": [ 0.1764 ], "steps": [ { "released_on": "2025-10-07", "cost_usd_per_task": "7.1432", "config": "GPT-5 pro (unknown)", "model_uid": "openai/gpt-5-pro-20251006" }, { "released_on": "2025-11-13", "cost_usd_per_task": "1.1704", "config": "GPT-5.1 (high)", "model_uid": "openai/gpt-5.1-20251113" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.79", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.7594", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 275, "ratio": 42.01882352941176, "t_half_days": 50.992368378120695, "t_half_ols_days": 62.307947342467635, "t_half_drop_first_days": 85.5070868574667, "t_half_drop_last_days": 31.530793333491317, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.1389, "q_count": 1, "q_values": [ 0.1389 ], "steps": [ { "released_on": "2025-10-07", "cost_usd_per_task": "7.1432", "config": "GPT-5 pro (unknown)", "model_uid": "openai/gpt-5-pro-20251006" }, { "released_on": "2025-11-13", "cost_usd_per_task": "1.1704", "config": "GPT-5.1 (high)", "model_uid": "openai/gpt-5.1-20251113" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 275, "ratio": 42.01882352941176, "t_half_days": 50.992368378120695, "t_half_ols_days": 63.82430239701622, "t_half_drop_first_days": 85.5070868574667, "t_half_drop_last_days": 31.530793333491317, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.1361, "q_count": 2, "q_values": [ 0.1361, 0.1319 ], "steps": [ { "released_on": "2025-09-29", "cost_usd_per_task": "0.7589", "config": "Claude Sonnet 4.5", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 283, "ratio": 4.464117647058823, "t_half_days": 131.11715696279134, "t_half_ols_days": 133.46056015799147, "t_half_drop_first_days": 151.58601109813728, "t_half_drop_last_days": 105.30196130078016, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.0986, "q_count": 1, "q_values": [ 0.0986 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.7302", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-03-05", "cost_usd_per_task": "0.27", "config": "GPT-5.4 (low)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 336, "ratio": 4.2952941176470585, "t_half_days": 159.79022416672598, "t_half_ols_days": 155.82588970976065, "t_half_drop_first_days": 151.58601109813728, "t_half_drop_last_days": 146.30758719310242, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.0972, "q_count": 1, "q_values": [ 0.0972 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.7302", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.2642", "config": "GPT-5.2 (low)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.17", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 336, "ratio": 4.2952941176470585, "t_half_days": 159.79022416672598, "t_half_ols_days": 163.6654123073969, "t_half_drop_first_days": 151.58601109813728, "t_half_drop_last_days": 85.90941098873094, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.0889, "q_count": 1, "q_values": [ 0.0889 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.7302", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.2642", "config": "GPT-5.2 (low)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 285, "ratio": 6.798882681564246, "t_half_days": 103.06304544724142, "t_half_ols_days": 100.92085405109297, "t_half_drop_first_days": 81.47995703258495, "t_half_drop_last_days": 85.90941098873094, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0861, "q_count": 1, "q_values": [ 0.0861 ], "steps": [ { "released_on": "2025-05-22", "cost_usd_per_task": "1.9284", "config": "Claude Opus 4", "model_uid": "anthropic/claude-opus-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.7302", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.48", "config": "Claude Opus 4.5", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.2642", "config": "GPT-5.2 (low)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 362, "ratio": 17.955307262569832, "t_half_days": 86.8868445338903, "t_half_ols_days": 89.82506927843983, "t_half_drop_first_days": 103.06304544724142, "t_half_drop_last_days": 70.78838826649108, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0778, "q_count": 1, "q_values": [ 0.0778 ], "steps": [ { "released_on": "2025-05-22", "cost_usd_per_task": "1.9284", "config": "Claude Opus 4", "model_uid": "anthropic/claude-opus-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.7302", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.22", "config": "Claude Opus 4.5 (none)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 362, "ratio": 17.955307262569832, "t_half_days": 86.8868445338903, "t_half_ols_days": 88.05808741715278, "t_half_drop_first_days": 103.06304544724142, "t_half_drop_last_days": 59.39021753218875, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0749, "q_count": 2, "q_values": [ 0.0749, 0.0736 ], "steps": [ { "released_on": "2025-05-22", "cost_usd_per_task": "1.9284", "config": "Claude Opus 4", "model_uid": "anthropic/claude-opus-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.22", "config": "Claude Opus 4.5 (none)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 362, "ratio": 17.955307262569832, "t_half_days": 86.8868445338903, "t_half_ols_days": 94.31715968574366, "t_half_drop_first_days": 138.18613112333642, "t_half_drop_last_days": 59.39021753218875, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0694, "q_count": 1, "q_values": [ 0.0694 ], "steps": [ { "released_on": "2025-05-22", "cost_usd_per_task": "1.9284", "config": "Claude Opus 4", "model_uid": "anthropic/claude-opus-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.2349", "config": "Claude Sonnet 4.5", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.22", "config": "Claude Opus 4.5 (none)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 362, "ratio": 17.955307262569832, "t_half_days": 86.8868445338903, "t_half_ols_days": 96.80208696797091, "t_half_drop_first_days": 138.18613112333642, "t_half_drop_last_days": 59.39021753218875, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0653, "q_count": 2, "q_values": [ 0.0653, 0.0611 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.2349", "config": "Claude Sonnet 4.5", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.22", "config": "Claude Opus 4.5 (none)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 398, "ratio": 7.764432029795159, "t_half_days": 134.60131856487945, "t_half_ols_days": 134.67501489376073, "t_half_drop_first_days": 138.18613112333642, "t_half_drop_last_days": 115.48239941417657, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.0593, "q_count": 1, "q_values": [ 0.0593 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.2349", "config": "Claude Sonnet 4.5", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-11-24", "cost_usd_per_task": "0.22", "config": "Claude Opus 4.5 (none)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 398, "ratio": 7.764432029795159, "t_half_days": 134.60131856487945, "t_half_ols_days": 141.90916817428695, "t_half_drop_first_days": 166.27841628425105, "t_half_drop_last_days": 115.48239941417657, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.058300000000000005, "q_count": 2, "q_values": [ 0.058300000000000005, 0.0569 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.1424", "config": "Claude Sonnet 4.5", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2026-05-19", "cost_usd_per_task": "0.1074", "config": "Gemini 3.5 Flash (minimal)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 398, "ratio": 7.764432029795159, "t_half_days": 134.60131856487945, "t_half_ols_days": 139.45533914680746, "t_half_drop_first_days": 166.27841628425105, "t_half_drop_last_days": 65.09994774250353, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (minimal)", "availability": "on_sale" } }, { "q": 0.05280000000000001, "q_count": 4, "q_values": [ 0.05280000000000001, 0.05140000000000001, 0.0486, 0.0452 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.4486", "config": "GPT-5 (medium)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 156, "ratio": 13.670491803278688, "t_half_days": 41.346482767513734, "t_half_ols_days": 50.085199514493254, "t_half_drop_first_days": 40.09108241465374, "t_half_drop_last_days": 126.33535804643141, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.0444, "q_count": 1, "q_values": [ 0.0444 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1977", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 156, "ratio": 13.670491803278688, "t_half_days": 41.346482767513734, "t_half_ols_days": 43.79341517397883, "t_half_drop_first_days": 40.09108241465374, "t_half_drop_last_days": 54.416880532193225, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.0403, "q_count": 3, "q_values": [ 0.0403, 0.0375, 0.0361 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.8339", "config": "o3 (high)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0629", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 156, "ratio": 13.670491803278688, "t_half_days": 41.346482767513734, "t_half_ols_days": 37.25277460270008, "t_half_drop_first_days": 40.09108241465374, "t_half_drop_last_days": 30.30512324957497, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.0299, "q_count": 1, "q_values": [ 0.0299 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-05-22", "cost_usd_per_task": "0.4857", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0629", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 231, "ratio": 8.970491803278689, "t_half_days": 72.98146807241572, "t_half_ols_days": 63.997445674425364, "t_half_drop_first_days": 40.09108241465374, "t_half_drop_last_days": 60.238332809835065, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.0298, "q_count": 1, "q_values": [ 0.0298 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.4787", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0629", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 231, "ratio": 8.970491803278689, "t_half_days": 72.98146807241572, "t_half_ols_days": 62.60990912310124, "t_half_drop_first_days": 52.485655751316415, "t_half_drop_last_days": 60.238332809835065, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.0292, "q_count": 2, "q_values": [ 0.0292, 0.0278 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.4787", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-06-17", "cost_usd_per_task": "0.4439", "config": "Gemini 2.5 Pro", "model_uid": "google/gemini-2.5-pro" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0629", "config": "GPT-5 mini (medium)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.061", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" } ], "valid": true, "span_days": 231, "ratio": 8.970491803278689, "t_half_days": 72.98146807241572, "t_half_ols_days": 64.61191115595514, "t_half_drop_first_days": 52.485655751316415, "t_half_drop_last_days": 60.238332809835065, "stale_frontier": false, "holder": { "model_uid": "xai/grok-4-fast-reasoning", "display": "Grok 4 Fast (reasoning)", "availability": "on_sale" } }, { "q": 0.026100000000000005, "q_count": 1, "q_values": [ 0.026100000000000005 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.4787", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-06-17", "cost_usd_per_task": "0.4439", "config": "Gemini 2.5 Pro", "model_uid": "google/gemini-2.5-pro" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0295", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 188, "ratio": 18.549152542372884, "t_half_days": 44.620803469289825, "t_half_ols_days": 52.339757908583735, "t_half_drop_first_days": 28.107110851348914, "t_half_drop_last_days": 453.8925938098668, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.025400000000000002, "q_count": 1, "q_values": [ 0.025400000000000002 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.4787", "config": "o3 (medium)", "model_uid": "openai/o3-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.3191", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0295", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 188, "ratio": 18.549152542372884, "t_half_days": 44.620803469289825, "t_half_ols_days": 44.6048892763451, "t_half_drop_first_days": 28.107110851348914, "t_half_drop_last_days": 140.0920918822776, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.0236, "q_count": 1, "q_values": [ 0.0236 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.2311", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0295", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.0216, "q_count": 2, "q_values": [ 0.0216, 0.0212 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.5472", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.2311", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0302", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0295", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 188, "ratio": 18.549152542372884, "t_half_days": 44.620803469289825, "t_half_ols_days": 41.04337713578206, "t_half_drop_first_days": 38.05059206441389, "t_half_drop_last_days": 26.08000048173203, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.020800000000000003, "q_count": 6, "q_values": [ 0.020800000000000003, 0.020500000000000004, 0.0199, 0.0198, 0.0194, 0.0169 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.2843", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.2311", "config": "o4-mini (medium)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0302", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0295", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 188, "ratio": 9.63728813559322, "t_half_days": 57.51650042324215, "t_half_ols_days": 51.14653201107934, "t_half_drop_first_days": 38.05059206441389, "t_half_drop_last_days": 33.69612160278001, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.0167, "q_count": 2, "q_values": [ 0.0167, 0.015300000000000001 ], "steps": [ { "released_on": "2025-01-31", "cost_usd_per_task": "0.2843", "config": "o3-mini (medium)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.05", "config": "o4-mini (low)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-05-20", "cost_usd_per_task": "0.0302", "config": "Gemini 2.5 Flash Preview (05-20)", "model_uid": "google/gemini-2.5-flash-preview-05-20" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0094", "config": "GPT-5 mini (minimal)", "model_uid": "openai/gpt-5-mini-20250807" } ], "valid": true, "span_days": 188, "ratio": 30.24468085106383, "t_half_days": 38.22218429352733, "t_half_ols_days": 38.519394163418504, "t_half_drop_first_days": 46.86472048436762, "t_half_drop_last_days": 33.69612160278001, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-mini-20250807", "display": "GPT-5 mini (minimal)", "availability": "on_sale" } }, { "q": 0.013000000000000001, "q_count": 8, "q_values": [ 0.013000000000000001, 0.012700000000000001, 0.0125, 0.0112, 0.0111, 0.009, 0.0088, 0.0085 ], "steps": [ { "released_on": "2025-01-20", "cost_usd_per_task": "0.08", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.004", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } }, { "q": 0.0083, "q_count": 1, "q_values": [ 0.0083 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.1907", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.08", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.004", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } }, { "q": 0.008, "q_count": 4, "q_values": [ 0.008, 0.007000000000000001, 0.0042, 0.004 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.1907", "config": "o1-mini (unknown)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.04", "config": "Gemini 1.5 Pro 002", "model_uid": "google/gemini-1.5-pro-002" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.004", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } }, { "q": 0, "q_count": 1, "q_values": [ 0 ], "steps": [ { "released_on": "2024-07-18", "cost_usd_per_task": "0.01", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.004", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.0036", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.0025", "config": "GPT-5 nano (minimal)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": true, "span_days": 385, "ratio": 4, "t_half_days": 192.5, "t_half_ols_days": 192.19869890007678, "t_half_drop_first_days": 269.8828820663216, "t_half_drop_last_days": 183.18358559565772, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (minimal)", "availability": "on_sale" } } ] }, { "benchmark": "cursorbench", "domain": "coding", "points": 27, "thresholds": 25, "curves": 2, "duplicate_thresholds": 6, "rejected_staircases": 17, "zero_cost_excluded": 0, "newest_release": "2026-07-24", "in_pools": [ "gate_no_arc_mirrors", "gate_all" ], "median_days": 100.86662896691308, "iqr_days": [ 83.54625807032036, 118.18699986350582 ], "full_band_days": [ 66.22588717372763, 135.50737076009855 ], "grid": [ { "q": 0.7290000000000001, "q_count": 1, "q_values": [ 0.7290000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "18.02", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "availability": "on_sale" } }, { "q": 0.72, "q_count": 1, "q_values": [ 0.72 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "13.74", "config": "Claude Fable 5 (xhigh)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (xhigh)", "availability": "on_sale" } }, { "q": 0.706, "q_count": 1, "q_values": [ 0.706 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (high)", "availability": "on_sale" } }, { "q": 0.7000000000000001, "q_count": 1, "q_values": [ 0.7000000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-24", "cost_usd_per_task": "8.23", "config": "Claude Opus 5 (max)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "availability": "on_sale" } }, { "q": 0.693, "q_count": 1, "q_values": [ 0.693 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-24", "cost_usd_per_task": "7.35", "config": "Claude Opus 5 (xhigh)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (xhigh)", "availability": "on_sale" } }, { "q": 0.672, "q_count": 1, "q_values": [ 0.672 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "5.69", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (max)", "availability": "on_sale" } }, { "q": 0.667, "q_count": 1, "q_values": [ 0.667 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "5.69", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "3.91", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.648, "q_count": 1, "q_values": [ 0.648 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "11.02", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-06-09", "cost_usd_per_task": "10.81", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "5.69", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "3.91", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": true, "span_days": 99, "ratio": 2.8184143222506393, "t_half_days": 66.22588717372763, "t_half_ols_days": 69.14674701939559, "t_half_drop_first_days": 30.67221198463643, "t_half_drop_last_days": 88.08506224372042, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.643, "q_count": 2, "q_values": [ 0.643, 0.638 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "11.02", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "4.37", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-24", "cost_usd_per_task": "3.91", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.626, "q_count": 2, "q_values": [ 0.626, 0.621 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "11.02", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "3.59", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (high)", "availability": "on_sale" } }, { "q": 0.616, "q_count": 2, "q_values": [ 0.616, 0.612 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "7.11", "config": "Claude Opus 4.7 (xhigh)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "3.59", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (high)", "availability": "on_sale" } }, { "q": 0.594, "q_count": 1, "q_values": [ 0.594 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "5.01", "config": "Claude Opus 4.7 (high)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "3.59", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (high)", "availability": "on_sale" } }, { "q": 0.5920000000000001, "q_count": 4, "q_values": [ 0.5920000000000001, 0.584, 0.5660000000000001, 0.543 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "5.01", "config": "Claude Opus 4.7 (high)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.22", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (medium)", "availability": "on_sale" } }, { "q": 0.527, "q_count": 1, "q_values": [ 0.527 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "2.93", "config": "Claude Opus 4.7 (medium)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.22", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (medium)", "availability": "on_sale" } }, { "q": 0.498, "q_count": 1, "q_values": [ 0.498 ], "steps": [ { "released_on": "2026-04-16", "cost_usd_per_task": "2.93", "config": "Claude Opus 4.7 (medium)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.22", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "1.94", "config": "Gemini 3.5 Flash (unknown)", "model_uid": "google/gemini-3.5-flash" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (unknown)", "availability": "on_sale" } }, { "q": 0.49, "q_count": 1, "q_values": [ 0.49 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "3.09", "config": "Claude Sonnet 4.6 (max)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-16", "cost_usd_per_task": "2.93", "config": "Claude Opus 4.7 (medium)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.22", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-19", "cost_usd_per_task": "1.94", "config": "Gemini 3.5 Flash (unknown)", "model_uid": "google/gemini-3.5-flash" } ], "valid": true, "span_days": 91, "ratio": 1.592783505154639, "t_half_days": 135.50737076009855, "t_half_ols_days": 142.54690813227884, "t_half_drop_first_days": 55.47672889234075, "t_half_drop_last_days": 136.25487213512258, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash", "display": "Gemini 3.5 Flash (unknown)", "availability": "on_sale" } }, { "q": 0.488, "q_count": 1, "q_values": [ 0.488 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "3.06", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-16", "cost_usd_per_task": "2.93", "config": "Claude Opus 4.7 (medium)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.19", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (low)", "availability": "on_sale" } }, { "q": 0.483, "q_count": 1, "q_values": [ 0.483 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "3.06", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-16", "cost_usd_per_task": "1.87", "config": "Claude Opus 4.7 (low)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.19", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (low)", "availability": "on_sale" } }, { "q": 0.46, "q_count": 1, "q_values": [ 0.46 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "2.64", "config": "Claude Sonnet 4.6 (medium)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-16", "cost_usd_per_task": "1.87", "config": "Claude Opus 4.7 (low)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.19", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.5-20260423", "display": "GPT-5.5 (low)", "availability": "on_sale" } } ] }, { "benchmark": "deepswe", "domain": "coding", "points": 45, "thresholds": 43, "curves": 8, "duplicate_thresholds": 13, "rejected_staircases": 22, "zero_cost_excluded": 0, "newest_release": "2026-07-24", "in_pools": [ "baseline_2026_08_15", "gate_no_arc_mirrors", "gate_all" ], "median_days": 44.037154450048, "iqr_days": [ 42.45231953034376, 53.04382113150902 ], "full_band_days": [ 38.45339762513309, 56.88984519621859 ], "grid": [ { "q": 0.7364864864864865, "q_count": 1, "q_values": [ 0.7364864864864865 ], "steps": [ { "released_on": "2026-07-24", "cost_usd_per_task": "11.837583271396396", "config": "Claude Opus 5 (max)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "availability": "on_sale" } }, { "q": 0.7315436241610739, "q_count": 1, "q_values": [ 0.7315436241610739 ], "steps": [ { "released_on": "2026-07-24", "cost_usd_per_task": "9.07220232326622", "config": "Claude Opus 5 (xhigh)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (xhigh)", "availability": "on_sale" } }, { "q": 0.7282850779510023, "q_count": 1, "q_values": [ 0.7282850779510023 ], "steps": [ { "released_on": "2026-07-24", "cost_usd_per_task": "6.076084489977728", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.7266666666666667, "q_count": 1, "q_values": [ 0.7266666666666667 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "8.386436346666667", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "6.076084489977728", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.7073170731707318, "q_count": 1, "q_values": [ 0.7073170731707318 ], "steps": [ { "released_on": "2026-07-09", "cost_usd_per_task": "4.703663860310422", "config": "GPT-5.6 Sol (xhigh)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (xhigh)", "availability": "on_sale" } }, { "q": 0.6991150442477876, "q_count": 3, "q_values": [ 0.6991150442477876, 0.6972477064220184, 0.6962305986696231 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "13.414521495535714", "config": "Claude Fable 5 (xhigh)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "4.703663860310422", "config": "GPT-5.6 Sol (xhigh)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (xhigh)", "availability": "on_sale" } }, { "q": 0.6940133037694013, "q_count": 1, "q_values": [ 0.6940133037694013 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "13.414521495535714", "config": "Claude Fable 5 (xhigh)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.4698331108647453", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.6890380313199106, "q_count": 1, "q_values": [ 0.6890380313199106 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "13.414521495535714", "config": "Claude Fable 5 (xhigh)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.4698331108647453", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "3.28978764541387", "config": "Claude Opus 5 (medium)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (medium)", "availability": "on_sale" } }, { "q": 0.6860465116279071, "q_count": 1, "q_values": [ 0.6860465116279071 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "9.177638214788733", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.4698331108647453", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "3.28978764541387", "config": "Claude Opus 5 (medium)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (medium)", "availability": "on_sale" } }, { "q": 0.671875, "q_count": 1, "q_values": [ 0.671875 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "9.177638214788733", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.028116810267857", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.6703539823008849, "q_count": 1, "q_values": [ 0.6703539823008849 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "7.226236674778761", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.028116810267857", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.6536697247706422, "q_count": 1, "q_values": [ 0.6536697247706422 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "7.226236674778761", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "6.088186528935186", "config": "Claude Fable 5 (medium)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.028116810267857", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.6438053097345133, "q_count": 1, "q_values": [ 0.6438053097345133 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "5.100437004424779", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.028116810267857", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.6106194690265486, "q_count": 2, "q_values": [ 0.6106194690265486, 0.6017699115044248 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "5.100437004424779", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.8620318075221238", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.5958429561200924, "q_count": 2, "q_values": [ 0.5958429561200924, 0.5897435897435898 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "5.100437004424779", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "3.757873125874126", "config": "Claude Fable 5 (low)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.8620318075221238", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (medium)", "availability": "on_sale" } }, { "q": 0.5812917594654788, "q_count": 1, "q_values": [ 0.5812917594654788 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "5.100437004424779", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "3.757873125874126", "config": "Claude Fable 5 (low)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.8620318075221238", "config": "GPT-5.6 Sol (medium)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "1.6626223919821828", "config": "Claude Opus 5 (low)", "model_uid": "anthropic/claude-opus-5" } ], "valid": true, "span_days": 92, "ratio": 3.0677061905463843, "t_half_days": 56.88984519621859, "t_half_ols_days": 53.78894374480356, "t_half_drop_first_days": 38.25048041237617, "t_half_drop_last_days": 52.966715803980016, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (low)", "availability": "on_sale" } }, { "q": 0.5685840707964602, "q_count": 2, "q_values": [ 0.5685840707964602, 0.5436241610738255 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "5.100437004424779", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "3.757873125874126", "config": "Claude Fable 5 (low)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.5356192685840708", "config": "GPT-5.6 Luna (xhigh)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (xhigh)", "availability": "on_sale" } }, { "q": 0.5398230088495575, "q_count": 2, "q_values": [ 0.5398230088495575, 0.5384615384615384 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.5356192685840708", "config": "GPT-5.6 Luna (xhigh)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (xhigh)", "availability": "on_sale" } }, { "q": 0.5376106194690266, "q_count": 1, "q_values": [ 0.5376106194690266 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-08", "cost_usd_per_task": "2.4157481725663716", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.1343762765486725", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.5176991150442478, "q_count": 5, "q_values": [ 0.5176991150442478, 0.49667405764966743, 0.48672566371681414, 0.4855875831485588, 0.4823008849557522 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-08", "cost_usd_per_task": "2.4157481725663716", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.1343762765486725", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 4.9828772351931265, "t_half_days": 54.38115682196772, "t_half_ols_days": 74.7908705403341, "t_half_drop_first_days": 60.28872040041896, "t_half_drop_last_days": 101.92364766116856, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.45353982300884954, "q_count": 1, "q_values": [ 0.45353982300884954 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-08", "cost_usd_per_task": "2.4157481725663716", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.0742664646017699", "config": "GPT-5.6 Sol (low)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": true, "span_days": 126, "ratio": 5.261690568226838, "t_half_days": 52.59804256802279, "t_half_ols_days": 72.87593757468555, "t_half_drop_first_days": 56.795765401680875, "t_half_drop_last_days": 101.92364766116856, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (low)", "availability": "on_sale" } }, { "q": 0.4424778761061947, "q_count": 1, "q_values": [ 0.4424778761061947 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-08", "cost_usd_per_task": "2.4157481725663716", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.7779007499999999", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 126, "ratio": 7.266296792434669, "t_half_days": 44.037154450048, "t_half_ols_days": 63.271307555642274, "t_half_drop_first_days": 42.27450537598859, "t_half_drop_last_days": 101.92364766116856, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.4079822616407982, "q_count": 3, "q_values": [ 0.4079822616407982, 0.3977777777777778, 0.37389380530973454 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-28", "cost_usd_per_task": "2.293369216740577", "config": "Claude Opus 4.8 (low)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.7779007499999999", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 126, "ratio": 7.266296792434669, "t_half_days": 44.037154450048, "t_half_ols_days": 46.55562018798671, "t_half_drop_first_days": 42.27450537598859, "t_half_drop_last_days": 64.54539460428187, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (high)", "availability": "on_sale" } }, { "q": 0.3511111111111111, "q_count": 1, "q_values": [ 0.3511111111111111 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-28", "cost_usd_per_task": "2.293369216740577", "config": "Claude Opus 4.8 (low)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.5832493688888889", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 9.691322487541921, "t_half_days": 38.45339762513309, "t_half_ols_days": 40.89712674760692, "t_half_drop_first_days": 34.422607404275716, "t_half_drop_last_days": 64.54539460428187, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.3051224944320713, "q_count": 1, "q_values": [ 0.3051224944320713 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "5.6524577245575225", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-28", "cost_usd_per_task": "2.293369216740577", "config": "Claude Opus 4.8 (low)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-06-30", "cost_usd_per_task": "2.1865550832962137", "config": "Claude Sonnet 5 (low)", "model_uid": "anthropic/claude-sonnet-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.5832493688888889", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 126, "ratio": 9.691322487541921, "t_half_days": 38.45339762513309, "t_half_ols_days": 51.68195831267989, "t_half_drop_first_days": 34.422607404275716, "t_half_drop_last_days": 85.38783495144759, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.29933481152993346, "q_count": 1, "q_values": [ 0.29933481152993346 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "5.5223517987804875", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.749342734513274", "config": "GPT-5.5 (medium)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-05-28", "cost_usd_per_task": "2.293369216740577", "config": "Claude Opus 4.8 (low)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-06-30", "cost_usd_per_task": "2.1865550832962137", "config": "Claude Sonnet 5 (low)", "model_uid": "anthropic/claude-sonnet-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.5832493688888889", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 142, "ratio": 9.468251649034386, "t_half_days": 43.785293498747315, "t_half_ols_days": 58.756103052040196, "t_half_drop_first_days": 34.422607404275716, "t_half_drop_last_days": 99.50448978143332, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.26991150442477874, "q_count": 1, "q_values": [ 0.26991150442477874 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "5.5223517987804875", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.2002025442477877", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.5832493688888889", "config": "GPT-5.6 Terra (medium)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (medium)", "availability": "on_sale" } }, { "q": 0.24053452115812918, "q_count": 2, "q_values": [ 0.24053452115812918, 0.11751662971175167 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "5.5223517987804875", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.2002025442477877", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.4277452605790646", "config": "GPT-5.6 Terra (low)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (low)", "availability": "on_sale" } }, { "q": 0.11283185840707964, "q_count": 1, "q_values": [ 0.11283185840707964 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "5.5223517987804875", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.2002025442477877", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.21630978982300886", "config": "GPT-5.6 Luna (medium)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (medium)", "availability": "on_sale" } }, { "q": 0.015486725663716814, "q_count": 1, "q_values": [ 0.015486725663716814 ], "steps": [ { "released_on": "2026-02-17", "cost_usd_per_task": "5.5223517987804875", "config": "Claude Sonnet 4.6 (high)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.2002025442477877", "config": "GPT-5.5 (low)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.07240623938053097", "config": "GPT-5.6 Luna (low)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (low)", "availability": "on_sale" } } ] }, { "benchmark": "proofbench", "domain": "math", "points": 33, "thresholds": 27, "curves": 8, "duplicate_thresholds": 8, "rejected_staircases": 11, "zero_cost_excluded": 1, "newest_release": "2026-07-24", "in_pools": [ "baseline_2026_08_15", "gate_no_arc_mirrors", "gate_all" ], "median_days": 125.22803582017255, "iqr_days": [ 59.48147429944728, 126.35539620727153 ], "full_band_days": [ 44.473484344767265, 151.77507497984644 ], "grid": [ { "q": 0.78, "q_count": 1, "q_values": [ 0.78 ], "steps": [ { "released_on": "2026-07-24", "cost_usd_per_task": "2.10619", "config": "Claude Opus 5 (max)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "availability": "on_sale" } }, { "q": 0.77, "q_count": 2, "q_values": [ 0.77, 0.71 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "availability": "on_sale" } }, { "q": 0.6900000000000001, "q_count": 2, "q_values": [ 0.6900000000000001, 0.66 ], "steps": [ { "released_on": "2026-05-28", "cost_usd_per_task": "2.293129", "config": "Claude Opus 4.8 (max)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "availability": "on_sale" } }, { "q": 0.56, "q_count": 1, "q_values": [ 0.56 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "3.202015", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-05-28", "cost_usd_per_task": "2.293129", "config": "Claude Opus 4.8 (max)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "availability": "on_sale" } }, { "q": 0.54, "q_count": 1, "q_values": [ 0.54 ], "steps": [ { "released_on": "2026-03-05", "cost_usd_per_task": "3.202015", "config": "GPT-5.4 (xhigh)", "model_uid": "openai/gpt-5.4-20260305" }, { "released_on": "2026-04-16", "cost_usd_per_task": "1.807786", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 126, "ratio": 7.126278306348412, "t_half_days": 44.473484344767265, "t_half_ols_days": 52.612372223818696, "t_half_drop_first_days": 41.82449049279124, "t_half_drop_last_days": 99.04177061789667, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.5, "q_count": 1, "q_values": [ 0.5 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.703583", "config": "Claude Opus 4.6 (max)", "model_uid": "anthropic/claude-opus-4-6-20260205" }, { "released_on": "2026-04-16", "cost_usd_per_task": "1.807786", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.68179", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 154, "ratio": 6.01698770377789, "t_half_days": 59.48147429944728, "t_half_ols_days": 71.57934365543608, "t_half_drop_first_days": 41.82449049279124, "t_half_drop_last_days": 170.99189185579698, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.45, "q_count": 1, "q_values": [ 0.45 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.703583", "config": "Claude Opus 4.6 (max)", "model_uid": "anthropic/claude-opus-4-6-20260205" }, { "released_on": "2026-02-17", "cost_usd_per_task": "2.162108", "config": "Claude Sonnet 4.6 (max)", "model_uid": "anthropic/claude-sonnet-4-6" }, { "released_on": "2026-04-16", "cost_usd_per_task": "1.807786", "config": "Claude Opus 4.7 (max)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-04-23", "cost_usd_per_task": "1.68179", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-06-09", "cost_usd_per_task": "1.635455", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 154, "ratio": 6.01698770377789, "t_half_days": 59.48147429944728, "t_half_ols_days": 79.69157720744099, "t_half_drop_first_days": 62.648698285842045, "t_half_drop_last_days": 170.99189185579698, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.36, "q_count": 1, "q_values": [ 0.36 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.267043", "config": "Claude Opus 4.5 (high)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.3, "q_count": 2, "q_values": [ 0.3, 0.29 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.267043", "config": "Claude Opus 4.5 (high)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-07-08", "cost_usd_per_task": "0.484548", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.26, "q_count": 1, "q_values": [ 0.26 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.267043", "config": "Claude Opus 4.5 (high)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.525953", "config": "Gemini 3.1 Pro Preview (high)", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-07-08", "cost_usd_per_task": "0.484548", "config": "Grok 4.5 (high)", "model_uid": "xai/grok-4.5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.449325", "config": "GPT-5.6 Luna (max)", "model_uid": "openai/gpt-5.6-luna" } ], "valid": true, "span_days": 227, "ratio": 2.819880932509876, "t_half_days": 151.77507497984644, "t_half_ols_days": 183.70001340769508, "t_half_drop_first_days": 616.2662159799338, "t_half_drop_last_days": 162.97050601624596, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-luna", "display": "GPT-5.6 Luna (max)", "availability": "on_sale" } }, { "q": 0.21, "q_count": 1, "q_values": [ 0.21 ], "steps": [ { "released_on": "2025-11-24", "cost_usd_per_task": "1.267043", "config": "Claude Opus 4.5 (high)", "model_uid": "anthropic/claude-opus-4-5-20251101" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.525953", "config": "Gemini 3.1 Pro Preview (high)", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.2, "q_count": 1, "q_values": [ 0.2 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "1.229076", "config": "Gemini 3 Pro Preview (high)", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.525953", "config": "Gemini 3.1 Pro Preview (high)", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.19, "q_count": 1, "q_values": [ 0.19 ], "steps": [ { "released_on": "2025-09-29", "cost_usd_per_task": "0.65341", "config": "Claude Sonnet 4.5 (unknown)", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.525953", "config": "Gemini 3.1 Pro Preview (high)", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.18, "q_count": 1, "q_values": [ 0.18 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.852361", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.65341", "config": "Claude Sonnet 4.5 (unknown)", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2026-02-19", "cost_usd_per_task": "0.525953", "config": "Gemini 3.1 Pro Preview (high)", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": true, "span_days": 222, "ratio": 3.417084601168212, "t_half_days": 125.22803582017255, "t_half_ols_days": 163.40939249224263, "t_half_drop_first_days": 121.64487618122074, "t_half_drop_last_days": 281.3946319962701, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.15, "q_count": 1, "q_values": [ 0.15 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.852361", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.65341", "config": "Claude Sonnet 4.5 (unknown)", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.327286", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": true, "span_days": 222, "ratio": 3.417084601168212, "t_half_days": 125.22803582017255, "t_half_ols_days": 118.96026370460632, "t_half_drop_first_days": 121.64487618122074, "t_half_drop_last_days": 95.5889540923451, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.14, "q_count": 1, "q_values": [ 0.14 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.852361", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.65341", "config": "Claude Sonnet 4.5 (unknown)", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.327286", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.31815", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" } ], "valid": true, "span_days": 222, "ratio": 3.417084601168212, "t_half_days": 125.22803582017255, "t_half_ols_days": 126.53303654514534, "t_half_drop_first_days": 121.64487618122074, "t_half_drop_last_days": 136.4508400091104, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-mini-20260317", "display": "GPT-5.4 mini (xhigh)", "availability": "on_sale" } }, { "q": 0.13, "q_count": 1, "q_values": [ 0.13 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.852361", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-09-29", "cost_usd_per_task": "0.65341", "config": "Claude Sonnet 4.5 (unknown)", "model_uid": "anthropic/claude-sonnet-4-5-20250929" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.327286", "config": "Gemini 3 Flash Preview (high)", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-02-17", "cost_usd_per_task": "0.31815", "config": "Grok 4.20 (reasoning)", "model_uid": "xai/grok-4.20-0309-reasoning" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.249441", "config": "GPT-5.4 mini (xhigh)", "model_uid": "openai/gpt-5.4-mini-20260317" }, { "released_on": "2026-07-21", "cost_usd_per_task": "0.132788", "config": "Gemini 3.5 Flash-Lite (high)", "model_uid": "google/gemini-3.5-flash-lite" } ], "valid": true, "span_days": 348, "ratio": 6.418961050697353, "t_half_days": 129.73747736856848, "t_half_ols_days": 130.95049477356255, "t_half_drop_first_days": 128.32426411530594, "t_half_drop_last_days": 125.22803582017255, "stale_frontier": false, "holder": { "model_uid": "google/gemini-3.5-flash-lite", "display": "Gemini 3.5 Flash-Lite (high)", "availability": "on_sale" } }, { "q": 0.12, "q_count": 5, "q_values": [ 0.12, 0.11, 0.1, 0.09, 0.08 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.049218", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5-nano-20250807", "display": "GPT-5 nano (high)", "availability": "on_sale" } }, { "q": 0.05, "q_count": 2, "q_values": [ 0.05, 0.04 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.049218", "config": "GPT-5 nano (high)", "model_uid": "openai/gpt-5-nano-20250807" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.015894", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.4-nano-20260317", "display": "GPT-5.4 nano (high)", "availability": "on_sale" } } ] }, { "benchmark": "the-agent-company", "domain": "agentic", "points": 6, "thresholds": 6, "curves": 1, "duplicate_thresholds": 0, "rejected_staircases": 5, "zero_cost_excluded": 0, "newest_release": "2025-05-22", "in_pools": [ "gate_no_arc_mirrors", "gate_all" ], "median_days": 43.206841227354964, "iqr_days": [ 43.206841227354964, 43.206841227354964 ], "full_band_days": [ 43.206841227354964, 43.206841227354964 ], "grid": [ { "q": 0.432, "q_count": 1, "q_values": [ 0.432 ], "steps": [ { "released_on": "2025-05-22", "cost_usd_per_task": "1.63", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-sonnet-4-20250514", "display": "Claude Sonnet 4", "availability": "on_sale" } }, { "q": 0.393, "q_count": 1, "q_values": [ 0.393 ], "steps": [ { "released_on": "2025-05-06", "cost_usd_per_task": "4.23", "config": "Gemini 2.5 Pro Preview (05-06)", "model_uid": "google/gemini-2.5-pro-preview-05-06" }, { "released_on": "2025-05-22", "cost_usd_per_task": "1.63", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-sonnet-4-20250514", "display": "Claude Sonnet 4", "availability": "on_sale" } }, { "q": 0.344, "q_count": 1, "q_values": [ 0.344 ], "steps": [ { "released_on": "2024-10-22", "cost_usd_per_task": "6.34", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-05-06", "cost_usd_per_task": "4.23", "config": "Gemini 2.5 Pro Preview (05-06)", "model_uid": "google/gemini-2.5-pro-preview-05-06" }, { "released_on": "2025-05-22", "cost_usd_per_task": "1.63", "config": "Claude Sonnet 4", "model_uid": "anthropic/claude-sonnet-4-20250514" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-sonnet-4-20250514", "display": "Claude Sonnet 4", "availability": "on_sale" } }, { "q": 0.19, "q_count": 1, "q_values": [ 0.19 ], "steps": [ { "released_on": "2024-10-22", "cost_usd_per_task": "6.34", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.79", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } }, { "q": 0.167, "q_count": 1, "q_values": [ 0.167 ], "steps": [ { "released_on": "2024-10-22", "cost_usd_per_task": "6.34", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2024-11-20", "cost_usd_per_task": "1.29", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.79", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } }, { "q": 0.08, "q_count": 1, "q_values": [ 0.08 ], "steps": [ { "released_on": "2024-09-24", "cost_usd_per_task": "6.78", "config": "Gemini 1.5 Pro 002", "model_uid": "google/gemini-1.5-pro-002" }, { "released_on": "2024-10-22", "cost_usd_per_task": "6.34", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2024-11-20", "cost_usd_per_task": "1.29", "config": "GPT-4o (2024-11-20)", "model_uid": "openai/gpt-4o-20241120" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.79", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" } ], "valid": true, "span_days": 134, "ratio": 8.582278481012658, "t_half_days": 43.206841227354964, "t_half_ols_days": 40.223521321279875, "t_half_drop_first_days": 35.27972835020874, "t_half_drop_last_days": 23.81037708717732, "stale_frontier": false, "holder": { "model_uid": "google/gemini-2.0-flash-001", "display": "Gemini 2.0 Flash 001", "availability": "on_sale" } } ] }, { "benchmark": "weirdml", "domain": "coding", "points": 118, "thresholds": 114, "curves": 44, "duplicate_thresholds": 53, "rejected_staircases": 17, "zero_cost_excluded": 0, "newest_release": "2026-07-31", "in_pools": [ "baseline_2026_08_15", "gate_no_arc_mirrors", "gate_all" ], "median_days": 73.731033690307, "iqr_days": [ 62.581356273883344, 84.3493397960112 ], "full_band_days": [ 25.200916416662086, 159.72018878434463 ], "grid": [ { "q": 0.9194, "q_count": 1, "q_values": [ 0.9194 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "8.424409705882354", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" } ], "valid": false, "rejected_because": "1 step on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-fable-5", "display": "Claude Fable 5 (max)", "availability": "on_sale" } }, { "q": 0.9178000000000001, "q_count": 1, "q_values": [ 0.9178000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "8.424409705882354", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-24", "cost_usd_per_task": "4.7419545588235295", "config": "Claude Opus 5 (max)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (max)", "availability": "on_sale" } }, { "q": 0.9159, "q_count": 1, "q_values": [ 0.9159 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "8.424409705882354", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-24", "cost_usd_per_task": "2.1933491176470588", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.8943000000000001, "q_count": 1, "q_values": [ 0.8943000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "8.424409705882354", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "3.8707419117647057", "config": "GPT-5.6 Sol (max)", "model_uid": "openai/gpt-5.6-sol" }, { "released_on": "2026-07-24", "cost_usd_per_task": "2.1933491176470588", "config": "Claude Opus 5 (high)", "model_uid": "anthropic/claude-opus-5" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "anthropic/claude-opus-5", "display": "Claude Opus 5 (high)", "availability": "on_sale" } }, { "q": 0.8876000000000001, "q_count": 1, "q_values": [ 0.8876000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "8.424409705882354", "config": "Claude Fable 5 (max)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.25479875", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.8785, "q_count": 3, "q_values": [ 0.8785, 0.8697, 0.8634000000000001 ], "steps": [ { "released_on": "2026-06-09", "cost_usd_per_task": "3.1068966666666666", "config": "Claude Fable 5 (high)", "model_uid": "anthropic/claude-fable-5" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.25479875", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.8491, "q_count": 1, "q_values": [ 0.8491 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "2.568345317647059", "config": "GPT-5.5 (xhigh)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.25479875", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.8390000000000001, "q_count": 2, "q_values": [ 0.8390000000000001, 0.8289 ], "steps": [ { "released_on": "2026-04-23", "cost_usd_per_task": "2.221576529411765", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.25479875", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "2 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.793, "q_count": 1, "q_values": [ 0.793 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.35", "config": "GPT-5.3-Codex", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.221576529411765", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "1.25479875", "config": "GPT-5.6 Sol (high)", "model_uid": "openai/gpt-5.6-sol" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-sol", "display": "GPT-5.6 Sol (high)", "availability": "on_sale" } }, { "q": 0.7827, "q_count": 4, "q_values": [ 0.7827, 0.7795000000000001, 0.779, 0.777 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.35", "config": "GPT-5.3-Codex", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-04-23", "cost_usd_per_task": "2.221576529411765", "config": "GPT-5.5 (high)", "model_uid": "openai/gpt-5.5-20260423" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.7640000000000001, "q_count": 1, "q_values": [ 0.7640000000000001 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.35", "config": "GPT-5.3-Codex", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-04-16", "cost_usd_per_task": "0.6895180000000002", "config": "Claude Opus 4.7", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.764, "q_count": 3, "q_values": [ 0.764, 0.7604000000000001, 0.7545000000000001 ], "steps": [ { "released_on": "2026-02-05", "cost_usd_per_task": "2.35", "config": "GPT-5.3-Codex", "model_uid": "openai/gpt-5.3-codex" }, { "released_on": "2026-04-16", "cost_usd_per_task": "0.6895", "config": "Claude Opus 4.7 (unknown)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.722, "q_count": 1, "q_values": [ 0.722 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "2.05", "config": "GPT-5.2 (xhigh)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-04-16", "cost_usd_per_task": "0.6895", "config": "Claude Opus 4.7 (unknown)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.721, "q_count": 1, "q_values": [ 0.721 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "2.05", "config": "GPT-5.2 (xhigh)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-19", "cost_usd_per_task": "1.36", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-04-16", "cost_usd_per_task": "0.6895", "config": "Claude Opus 4.7 (unknown)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 6.572501077470948, "t_half_days": 77.30699326784053, "t_half_ols_days": 75.22898153359176, "t_half_drop_first_days": 65.9001780834769, "t_half_drop_last_days": 80.15259767683703, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.7045, "q_count": 1, "q_values": [ 0.7045 ], "steps": [ { "released_on": "2025-12-11", "cost_usd_per_task": "2.05", "config": "GPT-5.2 (xhigh)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-02-19", "cost_usd_per_task": "1.36", "config": "Gemini 3.1 Pro Preview", "model_uid": "google/gemini-3.1-pro-preview" }, { "released_on": "2026-04-16", "cost_usd_per_task": "0.6895", "config": "Claude Opus 4.7 (unknown)", "model_uid": "anthropic/claude-opus-4-7-20260416" }, { "released_on": "2026-05-28", "cost_usd_per_task": "0.4752000990099009", "config": "Claude Opus 4.8 (none)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": true, "span_days": 210, "ratio": 6.572501077470948, "t_half_days": 77.30699326784053, "t_half_ols_days": 75.1744186575213, "t_half_drop_first_days": 65.9001780834769, "t_half_drop_last_days": 79.65796886596019, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.6993, "q_count": 6, "q_values": [ 0.6993, 0.6878000000000001, 0.6715000000000001, 0.6607, 0.659, 0.637 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2026-05-28", "cost_usd_per_task": "0.4752000990099009", "config": "Claude Opus 4.8 (none)", "model_uid": "anthropic/claude-opus-4-8" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.634, "q_count": 1, "q_values": [ 0.634 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-5.6-terra", "display": "GPT-5.6 Terra (high)", "availability": "on_sale" } }, { "q": 0.6296, "q_count": 2, "q_values": [ 0.6296, 0.6264000000000001 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.3119056164215686", "config": "GPT-5.6 Terra (high)", "model_uid": "openai/gpt-5.6-terra" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 255, "ratio": 18.397587126742952, "t_half_days": 60.693409178742485, "t_half_ols_days": 94.31251315961131, "t_half_drop_first_days": 60.04772771448052, "t_half_drop_last_days": 309.29531986318375, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.616, "q_count": 1, "q_values": [ 0.616 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 255, "ratio": 18.397587126742952, "t_half_days": 60.693409178742485, "t_half_ols_days": 63.80064208127521, "t_half_drop_first_days": 60.04772771448052, "t_half_drop_last_days": 23.314236246766555, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.6086, "q_count": 1, "q_values": [ 0.6086 ], "steps": [ { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 255, "ratio": 18.397587126742952, "t_half_days": 60.693409178742485, "t_half_ols_days": 93.78862906049534, "t_half_drop_first_days": 60.04772771448052, "t_half_drop_last_days": 167.08435050856602, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.6077, "q_count": 1, "q_values": [ 0.6077 ], "steps": [ { "released_on": "2025-11-13", "cost_usd_per_task": "0.6913346832", "config": "GPT-5.1 (high)", "model_uid": "openai/gpt-5.1-20251113" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 260, "ratio": 24.190926206558526, "t_half_days": 56.56608076173687, "t_half_ols_days": 87.18945363555501, "t_half_drop_first_days": 60.693409178742485, "t_half_drop_last_days": 133.00141346770144, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.607, "q_count": 3, "q_values": [ 0.607, 0.6039, 0.603 ], "steps": [ { "released_on": "2025-08-07", "cost_usd_per_task": "0.5931804044", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 358, "ratio": 20.756348175093184, "t_half_days": 81.81958087964131, "t_half_ols_days": 104.92532431900554, "t_half_drop_first_days": 60.693409178742485, "t_half_drop_last_days": 214.2117646665747, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.5821, "q_count": 2, "q_values": [ 0.5821, 0.5744 ], "steps": [ { "released_on": "2025-06-10", "cost_usd_per_task": "2.540382703", "config": "o3-pro (high)", "model_uid": "openai/o3-pro-20250610" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.5931804044", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.028578264316831685", "config": "DeepSeek-V4 Flash (0731) (max)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 416, "ratio": 88.89212706678606, "t_half_days": 64.25718945977059, "t_half_ols_days": 89.36860881616697, "t_half_drop_first_days": 81.81958087964131, "t_half_drop_last_days": 107.44347950556319, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (max)", "availability": "on_sale" } }, { "q": 0.5705, "q_count": 3, "q_values": [ 0.5705, 0.561, 0.5453 ], "steps": [ { "released_on": "2025-06-10", "cost_usd_per_task": "2.540382703", "config": "o3-pro (high)", "model_uid": "openai/o3-pro-20250610" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.5931804044", "config": "GPT-5 (high)", "model_uid": "openai/gpt-5-20250807" }, { "released_on": "2025-11-18", "cost_usd_per_task": "0.5257711077", "config": "Gemini 3 Pro Preview", "model_uid": "google/gemini-3-pro-preview" }, { "released_on": "2025-12-11", "cost_usd_per_task": "0.416", "config": "GPT-5.2 (medium)", "model_uid": "openai/gpt-5.2-20251211" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 416, "ratio": 100.24226761921126, "t_half_days": 62.581356273883344, "t_half_ols_days": 87.37436080193801, "t_half_drop_first_days": 78.70131161564582, "t_half_drop_last_days": 107.44347950556319, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.5403, "q_count": 1, "q_values": [ 0.5403 ], "steps": [ { "released_on": "2025-06-10", "cost_usd_per_task": "2.540382703", "config": "o3-pro (high)", "model_uid": "openai/o3-pro-20250610" }, { "released_on": "2025-06-17", "cost_usd_per_task": "0.4096601321", "config": "Gemini 2.5 Pro", "model_uid": "google/gemini-2.5-pro" }, { "released_on": "2025-12-17", "cost_usd_per_task": "0.222", "config": "Gemini 3 Flash Preview", "model_uid": "google/gemini-3-flash-preview" }, { "released_on": "2026-07-09", "cost_usd_per_task": "0.19999024607843144", "config": "GPT-5.6 Luna (high)", "model_uid": "openai/gpt-5.6-luna" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 416, "ratio": 100.24226761921126, "t_half_days": 62.581356273883344, "t_half_ols_days": 102.23744970542913, "t_half_drop_first_days": 101.87305198217344, "t_half_drop_last_days": 107.44347950556319, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.5267, "q_count": 1, "q_values": [ 0.5267 ], "steps": [ { "released_on": "2025-06-10", "cost_usd_per_task": "2.540382703", "config": "o3-pro (high)", "model_uid": "openai/o3-pro-20250610" }, { "released_on": "2025-06-17", "cost_usd_per_task": "0.4096601321", "config": "Gemini 2.5 Pro", "model_uid": "google/gemini-2.5-pro" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 416, "ratio": 100.24226761921126, "t_half_days": 62.581356273883344, "t_half_ols_days": 85.61131816347559, "t_half_drop_first_days": 101.87305198217344, "t_half_drop_last_days": 12.796368361185275, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.5256, "q_count": 3, "q_values": [ 0.5256, 0.5242, 0.5226 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.5219, "q_count": 4, "q_values": [ 0.5219, 0.4989, 0.49600000000000005, 0.496 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-03-03", "cost_usd_per_task": "0.0664239769607843", "config": "Gemini 3.1 Flash-Lite", "model_uid": "google/gemini-3.1-flash-lite" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 7.721639003496284, "t_half_days": 159.72018878434463, "t_half_ols_days": 171.48241594714509, "t_half_drop_first_days": 169.28225350426482, "t_half_drop_last_days": 205.93319830851445, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.49229999999999996, "q_count": 3, "q_values": [ 0.49229999999999996, 0.489, 0.4771 ], "steps": [ { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-03-03", "cost_usd_per_task": "0.0664239769607843", "config": "Gemini 3.1 Flash-Lite", "model_uid": "google/gemini-3.1-flash-lite" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.047747187745098035", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 471, "ratio": 7.721639003496284, "t_half_days": 159.72018878434463, "t_half_ols_days": 170.3145499063702, "t_half_drop_first_days": 169.28225350426482, "t_half_drop_last_days": 164.6154240293011, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.4756, "q_count": 1, "q_values": [ 0.4756 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2026-03-03", "cost_usd_per_task": "0.0664239769607843", "config": "Gemini 3.1 Flash-Lite", "model_uid": "google/gemini-3.1-flash-lite" }, { "released_on": "2026-03-17", "cost_usd_per_task": "0.047747187745098035", "config": "GPT-5.4 nano (high)", "model_uid": "openai/gpt-5.4-nano-20260317" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 687, "ratio": 72.95116007269804, "t_half_days": 111.00592182390383, "t_half_ols_days": 121.00703095619156, "t_half_drop_first_days": 159.72018878434463, "t_half_drop_last_days": 104.4550129622009, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.4673, "q_count": 6, "q_values": [ 0.4673, 0.467, 0.46530000000000005, 0.4643, 0.4611, 0.4573 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-07-31", "cost_usd_per_task": "0.025342430526912183", "config": "DeepSeek-V4 Flash (0731) (high)", "model_uid": "deepseek/deepseek-v4-flash-0731" } ], "valid": true, "span_days": 687, "ratio": 72.95116007269804, "t_half_days": 111.00592182390383, "t_half_ols_days": 109.8299421129689, "t_half_drop_first_days": 159.72018878434463, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash-0731", "display": "DeepSeek-V4 Flash (0731) (high)", "availability": "on_sale" } }, { "q": 0.45630000000000004, "q_count": 1, "q_values": [ 0.45630000000000004 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.014616010092615771", "config": "DeepSeek-V4 Flash (max)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 126.48867196212606, "t_half_days": 84.3493397960112, "t_half_ols_days": 84.66191026061415, "t_half_drop_first_days": 99.65501546989738, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (max)", "availability": "on_sale" } }, { "q": 0.454, "q_count": 3, "q_values": [ 0.454, 0.441, 0.4388 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-10-15", "cost_usd_per_task": "0.08938744", "config": "Claude Haiku 4.5", "model_uid": "anthropic/claude-haiku-4-5-20251001" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.014616010092615771", "config": "DeepSeek-V4 Flash (max)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 126.48867196212606, "t_half_days": 84.3493397960112, "t_half_ols_days": 86.1175816709152, "t_half_drop_first_days": 99.65501546989738, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (max)", "availability": "on_sale" } }, { "q": 0.4382, "q_count": 1, "q_values": [ 0.4382 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-10-15", "cost_usd_per_task": "0.08938744", "config": "Claude Haiku 4.5", "model_uid": "anthropic/claude-haiku-4-5-20251001" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.014616010092615771", "config": "DeepSeek-V4 Flash (max)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 126.48867196212606, "t_half_days": 84.3493397960112, "t_half_ols_days": 79.97107764169304, "t_half_drop_first_days": 72.91434936755763, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (max)", "availability": "on_sale" } }, { "q": 0.4376, "q_count": 2, "q_values": [ 0.4376, 0.43710000000000004 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-04-16", "cost_usd_per_task": "0.1956851", "config": "o4-mini (high)", "model_uid": "openai/o4-mini-20250416" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-10-15", "cost_usd_per_task": "0.08938744", "config": "Claude Haiku 4.5", "model_uid": "anthropic/claude-haiku-4-5-20251001" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 73.17424237946584, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.437, "q_count": 2, "q_values": [ 0.437, 0.434 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-10-15", "cost_usd_per_task": "0.08938744", "config": "Claude Haiku 4.5", "model_uid": "anthropic/claude-haiku-4-5-20251001" }, { "released_on": "2025-12-01", "cost_usd_per_task": "0.0352105525490196", "config": "DeepSeek-V3.2-Speciale", "model_uid": "deepseek/deepseek-v3.2-speciale" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 77.86705814715818, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 77.87336016475182, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.4286, "q_count": 3, "q_values": [ 0.4286, 0.4276, 0.4191 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-08-07", "cost_usd_per_task": "0.1097665112", "config": "GPT-5 mini (high)", "model_uid": "openai/gpt-5-mini-20250807" }, { "released_on": "2025-09-19", "cost_usd_per_task": "0.01272623039", "config": "Grok 4 Fast (reasoning)", "model_uid": "xai/grok-4-fast-reasoning" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 68.87059823817245, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 51.791767044116405, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.4188, "q_count": 1, "q_values": [ 0.4188 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.01148501118", "config": "gpt-oss-120b (medium)", "model_uid": "openai/gpt-oss-120b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 67.1030800828257, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 44.60716039076688, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.4163, "q_count": 1, "q_values": [ 0.4163 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.07106829941", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.01148501118", "config": "gpt-oss-120b (medium)", "model_uid": "openai/gpt-oss-120b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 66.86605207324432, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 44.60716039076688, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.40950000000000003, "q_count": 2, "q_values": [ 0.40950000000000003, 0.4095 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-17", "cost_usd_per_task": "0.10585353676470587", "config": "Gemini 2.5 Flash Preview (04-17)", "model_uid": "google/gemini-2.5-flash-preview-04-17" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.07106829941", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.01148501118", "config": "gpt-oss-120b (medium)", "model_uid": "openai/gpt-oss-120b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 67.01229597089987, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 44.60716039076688, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.4093, "q_count": 1, "q_values": [ 0.4093 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-12-17", "cost_usd_per_task": "1.585624412", "config": "o1 (high)", "model_uid": "openai/o1-20241217" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-17", "cost_usd_per_task": "0.10585353676470587", "config": "Gemini 2.5 Flash Preview (04-17)", "model_uid": "google/gemini-2.5-flash-preview-04-17" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.07106829941", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.0108003425", "config": "gpt-oss-20b (high)", "model_uid": "openai/gpt-oss-20b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 66.84023043371761, "t_half_drop_first_days": 63.47376513354806, "t_half_drop_last_days": 44.07401991878834, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.3997, "q_count": 2, "q_values": [ 0.3997, 0.3937 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.2483238", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-17", "cost_usd_per_task": "0.10585353676470587", "config": "Gemini 2.5 Flash Preview (04-17)", "model_uid": "google/gemini-2.5-flash-preview-04-17" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.07106829941", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.0108003425", "config": "gpt-oss-20b (high)", "model_uid": "openai/gpt-oss-20b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 79.23554854679348, "t_half_drop_first_days": 107.8112743129967, "t_half_drop_last_days": 44.07401991878834, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.3904, "q_count": 6, "q_values": [ 0.3904, 0.39, 0.3836, 0.3806, 0.3796, 0.379 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.2483238", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.09702961905", "config": "GPT-4.1", "model_uid": "openai/gpt-4.1-20250414" }, { "released_on": "2025-05-28", "cost_usd_per_task": "0.07106829941", "config": "DeepSeek-R1 (May 2025)", "model_uid": "deepseek/deepseek-r1-0528" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.0108003425", "config": "gpt-oss-20b (high)", "model_uid": "openai/gpt-oss-20b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 79.26844751592492, "t_half_drop_first_days": 107.8112743129967, "t_half_drop_last_days": 44.07401991878834, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.3761, "q_count": 3, "q_values": [ 0.3761, 0.375, 0.3724 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.2483238", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.01660665882", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.0108003425", "config": "gpt-oss-20b (high)", "model_uid": "openai/gpt-oss-20b" }, { "released_on": "2026-04-24", "cost_usd_per_task": "0.007279538164705883", "config": "DeepSeek-V4 Flash (high)", "model_uid": "deepseek/deepseek-v4-flash" } ], "valid": true, "span_days": 589, "ratio": 253.96662043253343, "t_half_days": 73.731033690307, "t_half_ols_days": 79.46332960369004, "t_half_drop_first_days": 107.8112743129967, "t_half_drop_last_days": 44.07401991878834, "stale_frontier": false, "holder": { "model_uid": "deepseek/deepseek-v4-flash", "display": "DeepSeek-V4 Flash (high)", "availability": "on_sale" } }, { "q": 0.3683, "q_count": 1, "q_values": [ 0.3683 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.2483238", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-01-31", "cost_usd_per_task": "0.1744674606", "config": "o3-mini (high)", "model_uid": "openai/o3-mini-20250131" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.01660665882", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 327, "ratio": 485.7212646153589, "t_half_days": 36.64282335063001, "t_half_ols_days": 39.10491739427752, "t_half_drop_first_days": 47.61334727884662, "t_half_drop_last_days": 31.47682771977426, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.3649, "q_count": 1, "q_values": [ 0.3649 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "1.848759706", "config": "o1-preview", "model_uid": "openai/o1-preview-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.2483238", "config": "Claude 3.5 Sonnet v2", "model_uid": "anthropic/claude-3-5-sonnet-20241022" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.07850552", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.01660665882", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 327, "ratio": 485.7212646153589, "t_half_days": 36.64282335063001, "t_half_ols_days": 39.26808202948973, "t_half_drop_first_days": 47.61334727884662, "t_half_drop_last_days": 31.47682771977426, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.3632, "q_count": 1, "q_values": [ 0.3632 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.07850552", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.01660665882", "config": "GPT-4.1 mini", "model_uid": "openai/gpt-4.1-mini-20250414" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 327, "ratio": 29.235065230127784, "t_half_days": 67.15092125445899, "t_half_ols_days": 64.5957252912414, "t_half_drop_first_days": 45.11762046072044, "t_half_drop_last_days": 77.97995081911316, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.3608, "q_count": 4, "q_values": [ 0.3608, 0.3522, 0.3506, 0.3313 ], "steps": [ { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.07850552", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-03-24", "cost_usd_per_task": "0.01390399906", "config": "DeepSeek-V3 (Mar 2025)", "model_uid": "deepseek/deepseek-v3-0324" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 327, "ratio": 29.235065230127784, "t_half_days": 67.15092125445899, "t_half_ols_days": 63.28699757522656, "t_half_drop_first_days": 45.11762046072044, "t_half_drop_last_days": 64.32138909134773, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.3097, "q_count": 1, "q_values": [ 0.3097 ], "steps": [ { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2025-01-20", "cost_usd_per_task": "0.07850552", "config": "DeepSeek-R1", "model_uid": "deepseek/deepseek-r1" }, { "released_on": "2025-03-24", "cost_usd_per_task": "0.01390399906", "config": "DeepSeek-V3 (Mar 2025)", "model_uid": "deepseek/deepseek-v3-0324" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 411, "ratio": 64.52633046983915, "t_half_days": 68.36536491387517, "t_half_ols_days": 68.72164678628432, "t_half_drop_first_days": 67.15092125445899, "t_half_drop_last_days": 66.86387467140331, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.3073, "q_count": 2, "q_values": [ 0.3073, 0.2589 ], "steps": [ { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.05748017882", "config": "Claude 3.5 Haiku", "model_uid": "anthropic/claude-3-5-haiku-20241022" }, { "released_on": "2025-03-24", "cost_usd_per_task": "0.01390399906", "config": "DeepSeek-V3 (Mar 2025)", "model_uid": "deepseek/deepseek-v3-0324" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 411, "ratio": 64.52633046983915, "t_half_days": 68.36536491387517, "t_half_ols_days": 68.29586083087796, "t_half_drop_first_days": 67.15092125445899, "t_half_drop_last_days": 66.86387467140331, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.2577, "q_count": 2, "q_values": [ 0.2577, 0.2512 ], "steps": [ { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-10-22", "cost_usd_per_task": "0.05748017882", "config": "Claude 3.5 Haiku", "model_uid": "anthropic/claude-3-5-haiku-20241022" }, { "released_on": "2025-02-05", "cost_usd_per_task": "0.006734424118", "config": "Gemini 2.0 Flash 001", "model_uid": "google/gemini-2.0-flash-001" }, { "released_on": "2025-08-05", "cost_usd_per_task": "0.003806215294", "config": "gpt-oss-20b (medium)", "model_uid": "openai/gpt-oss-20b" } ], "valid": true, "span_days": 411, "ratio": 64.52633046983915, "t_half_days": 68.36536491387517, "t_half_ols_days": 63.81654747373611, "t_half_drop_first_days": 67.15092125445899, "t_half_drop_last_days": 44.327790273626796, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-oss-20b", "display": "gpt-oss-20b (medium)", "availability": "on_sale" } }, { "q": 0.2487, "q_count": 1, "q_values": [ 0.2487 ], "steps": [ { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" } ], "valid": false, "rejected_because": "3 steps on the staircase, the gate needs 4", "span_days": null, "ratio": null, "t_half_days": null, "t_half_ols_days": null, "t_half_drop_first_days": null, "t_half_drop_last_days": null, "stale_frontier": true, "holder": { "model_uid": "google/gemini-1.5-flash-002", "display": "Gemini 1.5 Flash 002", "availability": "retired" } }, { "q": 0.2318, "q_count": 3, "q_values": [ 0.2318, 0.2224, 0.222 ], "steps": [ { "released_on": "2024-02-29", "cost_usd_per_task": "0.8497747059", "config": "Claude 3 Opus", "model_uid": "anthropic/claude-3-opus-20240229" }, { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" } ], "valid": true, "span_days": 208, "ratio": 305.2120481942718, "t_half_days": 25.200916416662086, "t_half_ols_days": 33.80985942586583, "t_half_drop_first_days": 14.854001967453131, "t_half_drop_last_days": 66.82688029625089, "stale_frontier": true, "holder": { "model_uid": "google/gemini-1.5-flash-002", "display": "Gemini 1.5 Flash 002", "availability": "retired" } }, { "q": 0.1898, "q_count": 1, "q_values": [ 0.1898 ], "steps": [ { "released_on": "2024-02-29", "cost_usd_per_task": "0.8497747059", "config": "Claude 3 Opus", "model_uid": "anthropic/claude-3-opus-20240229" }, { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 410, "ratio": 329.15092910034593, "t_half_days": 49.027782306369296, "t_half_ols_days": 46.13668503631145, "t_half_drop_first_days": 45.344972501378614, "t_half_drop_last_days": 25.200916416662086, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.1801, "q_count": 1, "q_values": [ 0.1801 ], "steps": [ { "released_on": "2024-02-29", "cost_usd_per_task": "0.8497747059", "config": "Claude 3 Opus", "model_uid": "anthropic/claude-3-opus-20240229" }, { "released_on": "2024-04-09", "cost_usd_per_task": "0.3210278313", "config": "GPT-4 Turbo", "model_uid": "openai/gpt-4-turbo-20240409" }, { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 410, "ratio": 329.15092910034593, "t_half_days": 49.027782306369296, "t_half_ols_days": 46.82041137360772, "t_half_drop_first_days": 53.17449252341179, "t_half_drop_last_days": 25.200916416662086, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.1244, "q_count": 1, "q_values": [ 0.1244 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2024-04-09", "cost_usd_per_task": "0.3210278313", "config": "GPT-4 Turbo", "model_uid": "openai/gpt-4-turbo-20240409" }, { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-09-12", "cost_usd_per_task": "0.1112749524", "config": "o1-mini (medium)", "model_uid": "openai/o1-mini-20240912" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 79.16443751644005, "t_half_drop_first_days": 53.17449252341179, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.1176, "q_count": 1, "q_values": [ 0.1176 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2024-04-09", "cost_usd_per_task": "0.3210278313", "config": "GPT-4 Turbo", "model_uid": "openai/gpt-4-turbo-20240409" }, { "released_on": "2024-06-20", "cost_usd_per_task": "0.2456011059", "config": "Claude 3.5 Sonnet", "model_uid": "anthropic/claude-3-5-sonnet-20240620" }, { "released_on": "2024-07-18", "cost_usd_per_task": "0.0052607625", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 72.76262694754016, "t_half_drop_first_days": 53.17449252341179, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.1016, "q_count": 1, "q_values": [ 0.1016 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2024-02-29", "cost_usd_per_task": "0.1782906071", "config": "Claude 3 Sonnet", "model_uid": "anthropic/claude-3-sonnet-20240229" }, { "released_on": "2024-07-18", "cost_usd_per_task": "0.0052607625", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 73.03748026014193, "t_half_drop_first_days": 67.10579215638397, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.0984, "q_count": 1, "q_values": [ 0.0984 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2024-02-29", "cost_usd_per_task": "0.1782906071", "config": "Claude 3 Sonnet", "model_uid": "anthropic/claude-3-sonnet-20240229" }, { "released_on": "2024-03-07", "cost_usd_per_task": "0.01407510882", "config": "Claude 3 Haiku", "model_uid": "anthropic/claude-3-haiku-20240307" }, { "released_on": "2024-07-18", "cost_usd_per_task": "0.0052607625", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 76.1341734753747, "t_half_drop_first_days": 67.10579215638397, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.0706, "q_count": 1, "q_values": [ 0.0706 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2023-11-21", "cost_usd_per_task": "0.1724719012", "config": "Claude 2.1", "model_uid": "anthropic/claude-2.1" }, { "released_on": "2024-03-07", "cost_usd_per_task": "0.01407510882", "config": "Claude 3 Haiku", "model_uid": "anthropic/claude-3-haiku-20240307" }, { "released_on": "2024-07-18", "cost_usd_per_task": "0.0052607625", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 76.57201773052648, "t_half_drop_first_days": 84.13222658755707, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } }, { "q": 0.0348, "q_count": 1, "q_values": [ 0.0348 ], "steps": [ { "released_on": "2023-06-13", "cost_usd_per_task": "0.6379120588", "config": "GPT-4 (0613)", "model_uid": "openai/gpt-4-0613" }, { "released_on": "2023-11-21", "cost_usd_per_task": "0.1724719012", "config": "Claude 2.1", "model_uid": "anthropic/claude-2.1" }, { "released_on": "2024-01-25", "cost_usd_per_task": "0.04237081395", "config": "GPT-3.5 Turbo (0125)", "model_uid": "openai/gpt-3.5-turbo-0125" }, { "released_on": "2024-03-07", "cost_usd_per_task": "0.01407510882", "config": "Claude 3 Haiku", "model_uid": "anthropic/claude-3-haiku-20240307" }, { "released_on": "2024-07-18", "cost_usd_per_task": "0.0052607625", "config": "GPT-4o mini", "model_uid": "openai/gpt-4o-mini-20240718" }, { "released_on": "2024-09-24", "cost_usd_per_task": "0.002784210882", "config": "Gemini 1.5 Flash 002", "model_uid": "google/gemini-1.5-flash-002" }, { "released_on": "2025-04-14", "cost_usd_per_task": "0.002581717476", "config": "GPT-4.1 nano", "model_uid": "openai/gpt-4.1-nano-20250414" } ], "valid": true, "span_days": 671, "ratio": 247.0882521926268, "t_half_days": 84.41438035919711, "t_half_ols_days": 76.98437097237135, "t_half_drop_first_days": 84.13222658755707, "t_half_drop_last_days": 59.821846313226416, "stale_frontier": false, "holder": { "model_uid": "openai/gpt-4.1-nano-20250414", "display": "GPT-4.1 nano", "availability": "on_sale" } } ] } ] }, "f2": { "note": "a KICK is the moment a configuration became dominated: something at least as strong arrived at a strictly lower cost, or something strictly stronger arrived at no higher cost. Overlapping score intervals suppress the event, and a victim is kicked once.", "newest_kick": "2026-07-31", "window_note": "the window ends on the NEWEST KICK, the corpus's own now — a wall-clock anchor would make the same database print a different frequency every quiet week. Frequency is counted in EVENT-DAYS, not kicks: twelve kicks in one afternoon are one day the channel spoke.", "frequencies": [ { "label": "12 mo, ARC excluded", "days": 365, "kicks": 98, "event_days": 19, "per_month": 1.5845479452054796, "silent_pct": 94.79452054794521, "display": "98 kicks · 19 event-days / 365 d = 1.58/month · silent 94.8%" }, { "label": "12 mo, ARC included", "days": 365, "kicks": 164, "event_days": 23, "per_month": 1.91813698630137, "silent_pct": 93.6986301369863, "display": "164 kicks · 23 event-days / 365 d = 1.92/month · silent 93.7%" }, { "label": "6 mo, ARC excluded", "days": 182, "kicks": 74, "event_days": 12, "per_month": 2.007032967032967, "silent_pct": 93.4065934065934, "display": "74 kicks · 12 event-days / 182 d = 2.01/month · silent 93.4%" } ], "benchmarks": [ { "benchmark": "aider-polyglot", "configs": 46, "kicks": 12, "same_model": 0, "born_dominated": 26, "ci_suppressed": 0, "event_days": 7 }, { "benchmark": "ale-bench", "configs": 66, "kicks": 21, "same_model": 0, "born_dominated": 33, "ci_suppressed": 0, "event_days": 9 }, { "benchmark": "arc-agi", "configs": 138, "kicks": 36, "same_model": 0, "born_dominated": 82, "ci_suppressed": 0, "event_days": 13 }, { "benchmark": "arc-agi-2", "configs": 133, "kicks": 45, "same_model": 0, "born_dominated": 74, "ci_suppressed": 0, "event_days": 14 }, { "benchmark": "critpt", "configs": 10, "kicks": 0, "same_model": 0, "born_dominated": 4, "ci_suppressed": 0, "event_days": 0 }, { "benchmark": "cursorbench", "configs": 27, "kicks": 9, "same_model": 0, "born_dominated": 7, "ci_suppressed": 0, "event_days": 4 }, { "benchmark": "deepswe", "configs": 45, "kicks": 12, "same_model": 0, "born_dominated": 10, "ci_suppressed": 31, "event_days": 3 }, { "benchmark": "exploitbench", "configs": 14, "kicks": 4, "same_model": 0, "born_dominated": 3, "ci_suppressed": 0, "event_days": 2 }, { "benchmark": "osworld-2", "configs": 4, "kicks": 0, "same_model": 0, "born_dominated": 1, "ci_suppressed": 0, "event_days": 0 }, { "benchmark": "proofbench", "configs": 33, "kicks": 13, "same_model": 0, "born_dominated": 1, "ci_suppressed": 36, "event_days": 4 }, { "benchmark": "surface-evolver-bench", "configs": 14, "kicks": 4, "same_model": 0, "born_dominated": 4, "ci_suppressed": 0, "event_days": 2 }, { "benchmark": "the-agent-company", "configs": 6, "kicks": 4, "same_model": 0, "born_dominated": 0, "ci_suppressed": 0, "event_days": 4 }, { "benchmark": "weirdml", "configs": 118, "kicks": 51, "same_model": 0, "born_dominated": 55, "ci_suppressed": 0, "event_days": 24 } ], "zero_cost_excluded": 2, "zero_cost_note": "counted over the cost F2 compares, which is per-task where the benchmark publishes one and the suite total otherwise. That is a WIDER set than F1's y axis, so this figure can exceed f1.corpus.zero_cost_excluded over the same corpus; the two are not a contradiction, they are two scopes.", "dominance": { "benchmark": "ale-bench", "measurer": "Epoch AI", "configs": 66, "dominated": 54, "display": "54 of 66 configs are dominated (82%)", "note": "dominated is not the same as bad: a dominated configuration can still be the right purchase for a reason this archive cannot see." } }, "saturation": { "decision": "S28 (19 Aug 2026): NO saturation label is published in Phase 1a and no curve is suppressed. The evidence is printed instead, because the threshold that would separate 'saturating' from 'saturated' has not been measured, and a guessed threshold whose documented consequence is DO-NOT-PLOT would delete benchmarks from charts on the strength of a number nobody chose.", "review": "revisited at T+90 (≈2026-11-13); until then a threshold is set only if it derives from a measurement.", "rows": [ { "benchmark": "aider-polyglot", "max_score": 88, "score_max": 100, "headroom_pct": 12, "highest_valid_q": null, "highest_valid_q_note": "not measured: no threshold produced a staircase past the gate", "saturation_state": null, "display": { "max_score": "88", "score_max": "100", "headroom": "12.0%", "highest_valid_q": "—" } }, { "benchmark": "ale-bench", "max_score": 2176.88, "score_max": null, "score_max_note": "unbounded above (a rating, not a fraction)", "headroom_pct": null, "headroom_note": "not defined: headroom needs a ceiling, and this benchmark has none", "highest_valid_q": 1244.92, "saturation_state": null, "display": { "max_score": "2176.88", "score_max": "∞", "headroom": "—", "highest_valid_q": "1244.92" } }, { "benchmark": "arc-agi", "max_score": 0.98, "score_max": 1, "headroom_pct": 2.0000000000000018, "highest_valid_q": 0.905, "saturation_state": null, "display": { "max_score": "0.98", "score_max": "1", "headroom": "2.0%", "highest_valid_q": "0.905" } }, { "benchmark": "arc-agi-2", "max_score": 0.925, "score_max": 1, "headroom_pct": 7.499999999999996, "highest_valid_q": 0.7208, "saturation_state": null, "display": { "max_score": "0.925", "score_max": "1", "headroom": "7.5%", "highest_valid_q": "0.7208" } }, { "benchmark": "critpt", "max_score": 0.323, "score_max": 1, "headroom_pct": 67.7, "highest_valid_q": null, "highest_valid_q_note": "not measured: no threshold produced a staircase past the gate", "saturation_state": null, "display": { "max_score": "0.323", "score_max": "1", "headroom": "67.7%", "highest_valid_q": "—" } }, { "benchmark": "cursorbench", "max_score": 0.7290000000000001, "score_max": 1, "headroom_pct": 27.09999999999999, "highest_valid_q": 0.648, "saturation_state": null, "display": { "max_score": "0.729", "score_max": "1", "headroom": "27.1%", "highest_valid_q": "0.648" } }, { "benchmark": "deepswe", "max_score": 0.7364864864864865, "score_max": 1, "headroom_pct": 26.351351351351347, "highest_valid_q": 0.5812917594654788, "saturation_state": null, "display": { "max_score": "0.7364864864864865", "score_max": "1", "headroom": "26.4%", "highest_valid_q": "0.5812917594654788" } }, { "benchmark": "exploitbench", "max_score": 0.738, "score_max": 1, "headroom_pct": 26.200000000000003, "highest_valid_q": null, "highest_valid_q_note": "not measured: no threshold produced a staircase past the gate", "saturation_state": null, "display": { "max_score": "0.738", "score_max": "1", "headroom": "26.2%", "highest_valid_q": "—" } }, { "benchmark": "osworld-2", "max_score": 0.20600000000000002, "score_max": 1, "headroom_pct": 79.4, "highest_valid_q": null, "highest_valid_q_note": "not measured: no threshold produced a staircase past the gate", "saturation_state": null, "display": { "max_score": "0.206", "score_max": "1", "headroom": "79.4%", "highest_valid_q": "—" } }, { "benchmark": "proofbench", "max_score": 0.78, "score_max": 1, "headroom_pct": 21.999999999999996, "highest_valid_q": 0.54, "saturation_state": null, "display": { "max_score": "0.78", "score_max": "1", "headroom": "22.0%", "highest_valid_q": "0.54" } }, { "benchmark": "surface-evolver-bench", "max_score": 0.9500000000000001, "score_max": 1, "headroom_pct": 4.999999999999993, "highest_valid_q": null, "highest_valid_q_note": "not measured: no threshold produced a staircase past the gate", "saturation_state": null, "display": { "max_score": "0.95", "score_max": "1", "headroom": "5.0%", "highest_valid_q": "—" } }, { "benchmark": "the-agent-company", "max_score": 0.518, "score_max": 1, "headroom_pct": 48.199999999999996, "highest_valid_q": 0.08, "saturation_state": null, "display": { "max_score": "0.518", "score_max": "1", "headroom": "48.2%", "highest_valid_q": "0.08" } }, { "benchmark": "weirdml", "max_score": 0.9194, "score_max": 1, "headroom_pct": 8.06, "highest_valid_q": 0.721, "saturation_state": null, "display": { "max_score": "0.9194", "score_max": "1", "headroom": "8.1%", "highest_valid_q": "0.721" } } ] }, "suspect": { "decision": "S31 (19 Aug 2026): price_fact is NOT corrected, filtered or deleted — the archive records what the catalogues actually published. Doubt is this separate, derived claim, and it is presented per INCIDENT (seller × upstream commit), never by blaming the seller whose arrival made an old broken row visible again.", "rows_examined": 6702, "counts": { "magnitude": 15, "residue": 9, "curated": 10, "both_magnitude_residue": 3, "both_magnitude_curated": 0, "both_residue_curated": 0, "all_three": 0, "union": 31 }, "share_pct": "0.46%", "identity": { "claim": "union == M + R + C − MR − MC − RC + MRC, with each rule set built in its own pass over the ledger so the equation can actually fail", "terms": "31 == 15 + 9 + 10 − 3 − 0 − 0 + 0", "holds": true }, "rules": [ { "id": "m", "label": "magnitude two-tail", "definition": "1 ≤ price_micro ≤ 999 (a nonzero price under $0.001/Mtok) or price_micro ≥ 1e9 (at or above $1000/Mtok)", "rows": 15, "note": "measured: 15 of 15 independently confirmed wrong, and the nearest legitimate row sits 2.8× above the top of the bad band." }, { "id": "r", "label": "residue ratio", "definition": "the parser recorded what rounding threw away: residue ≠ 0 and (price_micro = 0 or 1e6·|residue| > price_micro), in exact rational arithmetic", "rows": 9, "note": "a WITNESS, not a detector: its recall over the classified upstream errors is 3 of 24, because an over-scale by a power of ten on a round decimal lands on a whole µUSD and leaves no residue at all. That blindness is why the third rule is a hand-curated file." }, { "id": "c", "label": "curated incident", "definition": "an entry in suspects.toml, each carrying a written rationale and its evidence, each required to select exactly one price_fact row", "rows": 10, "note": "a stale entry marks nothing while the count still claims it did, so a curated entry that selects zero or several rows fails the export outright." } ], "coverage": { "source_refs": 7349, "cited_observations": 2896, "rows_citing_nothing": 0, "dangling_source_refs": 0, "matched_pairs": 7325, "micro_fallback_pairs": 0, "ambiguous_micro_pairs": 0, "rows_without_comparable_entry": 16, "rows_with_residue": 71, "rows_residue_cleared": 62, "rows_with_several_firing_readings": 0, "rows_with_unreadable_residue": 0 }, "incidents": [ { "incident": "helicone-2025-10-23-mistral-off-band", "rows": [ { "seller": "helicone", "model_uid": "mistral/mistral-nemo", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 20000000, "price_usd_per_unit": "20", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$20/Mtok against a five-seller consensus of $0.15 — 133x, not a clean power of ten, in a catalogue whose other 94 comparable rows sit at median ratio 1.00. No correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 25cb43927a99." }, { "seller": "helicone", "model_uid": "mistral/mistral-nemo", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 40000000, "price_usd_per_unit": "40", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$40/Mtok against a consensus of $0.15 — 267x, log10 closeness 0.426, so not a decimal shift. No correction observed through 2026-08-17, the archive's as-of date, and the row was still driving NETSPLIT events in the archive's final fortnight.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT fc06d4725f67, 3cfc8d37f133, 330f15318621." }, { "seller": "helicone", "model_uid": "mistral/mistral-small", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 75000000, "price_usd_per_unit": "75", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$75/Mtok against a consensus of $1.00 — 75x, log10 closeness 0.125. No correction observed through 2026-08-17, the archive's as-of date. Note the consensus side of this model carries its own identity smear (11 floating aliases across two namespaces), which is why the entry names the row and not a corrected price.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 5a80a91c2525, afca45144ce9." }, { "seller": "helicone", "model_uid": "mistral/mistral-small", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 200000000, "price_usd_per_unit": "200", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$200/Mtok against a consensus of $3.00 — 67x, log10 closeness 0.176. No correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT b41cfc097a65, 2594c7fbf79f." } ], "row_count": 4 }, { "incident": "oci-2025-08-07-decimal-shift", "rows": [ { "seller": "oracle-oci", "model_uid": "xai/grok-3", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000, "price_usd_per_unit": "0.15", "valid_from": "2025-08-07T16:45:17Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "oci-2025-08-07-decimal-shift", "why": "$0.15/Mtok against a five-seller consensus of $15.00 — exactly 10^-2, while the same listing's input_text meter is correct at ratio 1.000. A two-cell decimal error in litellm provider 'oci'; no correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; oci book-wide 32 comparable era-rows, median ratio 1.00, only these two off-band." }, { "seller": "oracle-oci", "model_uid": "xai/grok-4-0709", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000, "price_usd_per_unit": "0.15", "valid_from": "2025-08-07T16:45:17Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "oci-2025-08-07-decimal-shift", "why": "$0.15/Mtok against a five-seller consensus of $15.00 — exactly 10^-2, same commit and same two-cell error as xai/grok-3; the listing's input_text meter is correct.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; still open at the 2026-08-17 archive as-of (374+ days), driver of NETSPLIT events 1b99d94cf7a2, 96c32ec44388, 4daa48899afb, cc4bad402b98." } ], "row_count": 2 }, { "incident": "watsonx-2025-10-06-bad-import", "rows": [ { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-small-2503", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 200000000, "price_usd_per_unit": "200", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$200/Mtok, 200x a three-seller consensus; corrected upstream 11 days later to $0.10/Mtok (÷2000). Part of one litellm commit that wrote 8 wrong rows over 4 models.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; window closes 2025-10-17T18:51:22Z, the incident's own correction commit." }, { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-small-2503", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 600000000, "price_usd_per_unit": "600", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$600/Mtok, 200x consensus; corrected upstream 11 days later to $0.30/Mtok (÷2000). Same commit as the input_text row.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; window closes 2025-10-17T18:51:22Z." }, { "seller": "ibm-watsonx", "model_uid": "mistral/pixtral-12b-2409", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000000, "price_usd_per_unit": "150", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$150/Mtok, exactly 1000x a four-seller consensus; corrected upstream 11 days later to $0.35/Mtok. The full Class-1 signature — single seller, clean power of ten, later corrected.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 8ed462c8dfd6." }, { "seller": "ibm-watsonx", "model_uid": "mistral/pixtral-12b-2409", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000000, "price_usd_per_unit": "150", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$150/Mtok, exactly 1000x consensus; corrected upstream 11 days later to $0.35/Mtok. Same commit as the input_text row.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT e1f9455d63b8." } ], "row_count": 4 } ], "rows": [ { "seller": "azure", "model_uid": "xai/grok-4-fast-non-reasoning", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 2500000000, "price_usd_per_unit": "2500", "valid_from": "2025-10-02T21:21:12Z", "valid_to": "2025-10-04T21:50:44Z", "confidence": "observed", "rules": "m" }, { "seller": "azure", "model_uid": "xai/grok-4-fast-reasoning", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 2900000000, "price_usd_per_unit": "2900", "valid_from": "2025-10-02T21:21:12Z", "valid_to": "2025-10-04T21:50:44Z", "confidence": "observed", "rules": "m" }, { "seller": "google-vertex", "model_uid": "google/gemini-1.5-flash", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 4688, "price_usd_per_unit": "0.004688", "valid_from": "2024-10-10T07:42:11Z", "valid_to": "2024-11-08T05:11:14Z", "confidence": "provisional", "rules": "r", "residue": { "observation_id": 1393, "field": "output_cost_per_token", "text": "-0.50000000000000", "ratio_exact": "1/9376" } }, { "seller": "google-vertex", "model_uid": "google/gemini-1.5-flash-001", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 4688, "price_usd_per_unit": "0.004688", "valid_from": "2024-10-10T07:42:11Z", "valid_to": "2024-11-08T05:11:14Z", "confidence": "provisional", "rules": "r", "residue": { "observation_id": 1394, "field": "output_cost_per_token", "text": "-0.50000000000000", "ratio_exact": "1/9376" } }, { "seller": "google-vertex", "model_uid": "google/gemini-1.5-flash-002", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 4688, "price_usd_per_unit": "0.004688", "valid_from": "2024-10-10T07:42:11Z", "valid_to": "2024-11-08T05:11:14Z", "confidence": "provisional", "rules": "r", "residue": { "observation_id": 1395, "field": "output_cost_per_token", "text": "-0.50000000000000", "ratio_exact": "1/9376" } }, { "seller": "helicone", "model_uid": "mistral/mistral-nemo", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 20000000, "price_usd_per_unit": "20", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$20/Mtok against a five-seller consensus of $0.15 — 133x, not a clean power of ten, in a catalogue whose other 94 comparable rows sit at median ratio 1.00. No correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 25cb43927a99." }, { "seller": "helicone", "model_uid": "mistral/mistral-nemo", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 40000000, "price_usd_per_unit": "40", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$40/Mtok against a consensus of $0.15 — 267x, log10 closeness 0.426, so not a decimal shift. No correction observed through 2026-08-17, the archive's as-of date, and the row was still driving NETSPLIT events in the archive's final fortnight.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT fc06d4725f67, 3cfc8d37f133, 330f15318621." }, { "seller": "helicone", "model_uid": "mistral/mistral-small", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 75000000, "price_usd_per_unit": "75", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$75/Mtok against a consensus of $1.00 — 75x, log10 closeness 0.125. No correction observed through 2026-08-17, the archive's as-of date. Note the consensus side of this model carries its own identity smear (11 floating aliases across two namespaces), which is why the entry names the row and not a corrected price.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 5a80a91c2525, afca45144ce9." }, { "seller": "helicone", "model_uid": "mistral/mistral-small", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 200000000, "price_usd_per_unit": "200", "valid_from": "2025-10-23T20:43:27Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "helicone-2025-10-23-mistral-off-band", "why": "$200/Mtok against a consensus of $3.00 — 67x, log10 closeness 0.176. No correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT b41cfc097a65, 2594c7fbf79f." }, { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-medium-2505", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 2250000000, "price_usd_per_unit": "2250", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "m" }, { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-medium-2505", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 6750000000, "price_usd_per_unit": "6750", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "m" }, { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-small-2503", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 200000000, "price_usd_per_unit": "200", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$200/Mtok, 200x a three-seller consensus; corrected upstream 11 days later to $0.10/Mtok (÷2000). Part of one litellm commit that wrote 8 wrong rows over 4 models.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; window closes 2025-10-17T18:51:22Z, the incident's own correction commit." }, { "seller": "ibm-watsonx", "model_uid": "mistral/mistral-small-2503", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 600000000, "price_usd_per_unit": "600", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$600/Mtok, 200x consensus; corrected upstream 11 days later to $0.30/Mtok (÷2000). Same commit as the input_text row.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; window closes 2025-10-17T18:51:22Z." }, { "seller": "ibm-watsonx", "model_uid": "mistral/pixtral-12b-2409", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000000, "price_usd_per_unit": "150", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$150/Mtok, exactly 1000x a four-seller consensus; corrected upstream 11 days later to $0.35/Mtok. The full Class-1 signature — single seller, clean power of ten, later corrected.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT 8ed462c8dfd6." }, { "seller": "ibm-watsonx", "model_uid": "mistral/pixtral-12b-2409", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000000, "price_usd_per_unit": "150", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:51:22Z", "confidence": "observed", "rules": "c", "incident": "watsonx-2025-10-06-bad-import", "why": "$150/Mtok, exactly 1000x consensus; corrected upstream 11 days later to $0.35/Mtok. Same commit as the input_text row.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; drives NETSPLIT e1f9455d63b8." }, { "seller": "ibm-watsonx", "model_uid": "openai/gpt-oss-120b", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 4000000000, "price_usd_per_unit": "4000", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:40:34Z", "confidence": "observed", "rules": "m" }, { "seller": "ibm-watsonx", "model_uid": "openai/gpt-oss-120b", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 16000000000, "price_usd_per_unit": "16000", "valid_from": "2025-10-06T19:33:05Z", "valid_to": "2025-10-17T18:40:34Z", "confidence": "observed", "rules": "m" }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-mini-20260317", "region": "unspecified", "service_tier": "standard", "meter": "cache_read", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 0, "price_usd_per_unit": "0", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "r", "residue": { "observation_id": 15186, "field": "cost.cache_read", "text": "0.07500000000000", "ratio_note": "undefined: price_micro is 0, so the ratio has no value — undefined, not infinite" } }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-mini-20260317", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 1, "price_usd_per_unit": "0.000001", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "m+r", "residue": { "observation_id": 15186, "field": "cost.input", "text": "-0.25000000000000", "ratio_exact": "1/4" } }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-mini-20260317", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 5, "price_usd_per_unit": "0.000005", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "m+r", "residue": { "observation_id": 15186, "field": "cost.output", "text": "-0.50000000000000", "ratio_exact": "1/10" } }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-nano-20260317", "region": "unspecified", "service_tier": "standard", "meter": "cache_read", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 0, "price_usd_per_unit": "0", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "r", "residue": { "observation_id": 15187, "field": "cost.cache_read", "text": "0.02000000000000", "ratio_note": "undefined: price_micro is 0, so the ratio has no value — undefined, not infinite" } }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-nano-20260317", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 0, "price_usd_per_unit": "0", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "r", "residue": { "observation_id": 15187, "field": "cost.input", "text": "0.20000000000000", "ratio_note": "undefined: price_micro is 0, so the ratio has no value — undefined, not infinite" } }, { "seller": "openrouter", "model_uid": "openai/gpt-5.4-nano-20260317", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 1, "price_usd_per_unit": "0.000001", "valid_from": "2026-03-18T21:16:55Z", "valid_to": "2026-04-21T22:41:02Z", "confidence": "observed", "rules": "m+r", "residue": { "observation_id": 15187, "field": "cost.output", "text": "0.25000000000000", "ratio_exact": "1/4" } }, { "seller": "oracle-oci", "model_uid": "xai/grok-3", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000, "price_usd_per_unit": "0.15", "valid_from": "2025-08-07T16:45:17Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "oci-2025-08-07-decimal-shift", "why": "$0.15/Mtok against a five-seller consensus of $15.00 — exactly 10^-2, while the same listing's input_text meter is correct at ratio 1.000. A two-cell decimal error in litellm provider 'oci'; no correction observed through 2026-08-17, the archive's as-of date.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; oci book-wide 32 comparable era-rows, median ratio 1.00, only these two off-band." }, { "seller": "oracle-oci", "model_uid": "xai/grok-4-0709", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 150000, "price_usd_per_unit": "0.15", "valid_from": "2025-08-07T16:45:17Z", "valid_to": "", "confidence": "observed", "rules": "c", "incident": "oci-2025-08-07-decimal-shift", "why": "$0.15/Mtok against a five-seller consensus of $15.00 — exactly 10^-2, same commit and same two-cell error as xai/grok-3; the listing's input_text meter is correct.", "evidence": "docs/DECISIONS.md 2026-08-19, S31 measurement round; still open at the 2026-08-17 archive as-of (374+ days), driver of NETSPLIT events 1b99d94cf7a2, 96c32ec44388, 4daa48899afb, cc4bad402b98." }, { "seller": "wandb", "model_uid": "deepseek/deepseek-v3.1", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 55000000000, "price_usd_per_unit": "55000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" }, { "seller": "wandb", "model_uid": "deepseek/deepseek-v3.1", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 165000000000, "price_usd_per_unit": "165000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" }, { "seller": "wandb", "model_uid": "openai/gpt-oss-120b", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 15000000000, "price_usd_per_unit": "15000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" }, { "seller": "wandb", "model_uid": "openai/gpt-oss-120b", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 60000000000, "price_usd_per_unit": "60000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" }, { "seller": "wandb", "model_uid": "openai/gpt-oss-20b", "region": "unspecified", "service_tier": "standard", "meter": "input_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 5000000000, "price_usd_per_unit": "5000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" }, { "seller": "wandb", "model_uid": "openai/gpt-oss-20b", "region": "unspecified", "service_tier": "standard", "meter": "output_text", "ctx_tier_min": 0, "time_window": "all", "unit": "mtok", "price_micro": 20000000000, "price_usd_per_unit": "20000", "valid_from": "2025-09-10T18:37:30Z", "valid_to": "", "confidence": "observed", "rules": "m" } ], "precision_note": "100% precision is a MEASURED PROPERTY of this corpus, not a guarantee. A future error with a mid-sized residue would land in the empty band between the benign artefacts and the real ones, and would want a re-measurement rather than a silent pass." }, "corrections": { "policy": "A correction is published when a number this archive PUBLISHED turns out to be wrong: the old value stays visible, the new one is dated, and the reason is written out. Nothing is silently edited. The list below is empty, and that is the honest state — this archive has not yet had to correct a published number. It is not a placeholder and it is not a claim that nothing will ever be wrong. Upstream errors this archive faithfully recorded live in the data-quality section instead, because those are the catalogues' numbers, not ours.", "entries": [], "count": 0 } }