SCROLLBACK TELEMETRY · LIVE MODEL INDEX · 2026-10-18
LLM Performance, Value / Cost & Effort Parity Telemetry
Verified benchmark scores, real per-task token spend, and unit cost of capability across currently purchasable frontier configurations. To resolve cross-vendor reasoning effort asymmetry, use the ⚖️ Effort Parity Filter to compare models within identical effort tiers, or inspect the benchmarks where frontier labs share effort tiers (DeepSWE, CursorBench, WeirdML).
CODING · ALE-BENCH ⚠️ Asymmetric Effort
Algorithmic Coding & Competitive Engineering
🏆 SOTA SCORE GPT-6 Astra MAX 2951.32
⚡ VALUE / COST GPT-6 Luna XHIGH $0.00001144
💎 ELITE PARETO GPT-6.1 Sol MAX $0.596828/task
REASONING · PROOFBENCH ℹ️ Effort Unclear
Formal Mathematical Proof & Deep Reasoning
🏆 SOTA SCORE Claude Sonnet 5.5 MAX 1
⚡ VALUE / COST GPT-6 Luna $0.06506
💎 ELITE PARETO Claude Haiku 5.5 MAX $0.173874/task
AGENT · DEEPSWE ✅ Effort Parity
Autonomous SWE Agent & Repository Resolution
🏆 SOTA SCORE GPT-6 Astra XHIGH 0.741
⚡ VALUE / COST GPT-5.6 Luna HIGH $1.758
💎 ELITE PARETO GPT-6 Astra LOW $1.59522/task
IDE · CURSORBENCH ✅ Effort Parity
IDE Code Editing & Multi-Step Completion
🏆 SOTA SCORE Claude Opus 5.5 MAX 0.578
⚡ VALUE / COST Claude Haiku 5.5 LOW $0.2589
💎 ELITE PARETO Claude Haiku 5.5 XHIGH $0.56/task
ML-CODE · WEIRDML ✅ Effort Parity
Algorithmic ML & Scientific Coding
🏆 SOTA SCORE GPT-6 Astra MAX 0.936
⚡ VALUE / COST DeepSeek-V4 Flash (0731) HIGH $0.04442
💎 ELITE PARETO GPT-5.6 Terra HIGH $0.311906/task
ABSTRACT · ARC-AGI-2 ⚠️ Asymmetric Effort
Abstract Fluid Intelligence & Pattern Discovery
🏆 SOTA SCORE GPT-5.6 Sol MAX 0.925
⚡ VALUE / COST GPT-5.6 Sol MEDIUM $0.7007
💎 ELITE PARETO GPT-5.6 Sol HIGH $0.74/task
CODING · ALE-BENCH ⚠️ ASYMMETRIC EFFORT (OPENAI MAX · ANTHROPIC HIGH)
Algorithmic Coding & Competitive Engineering — ALE-Bench
Metric: rating (Epoch AI) · Unit cost ratio: $ / rating point · Cost basis: list_price_x_tokens (community-submitted).
⚖️ Methodological Effort Alert: frontier labs measured at different effort tiers
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). Only 25% of one compared lab's rows sit at an effort another lab was also measured at; filter to one effort tier before comparing labs. OpenAI: 12 rows — MAX 5 (59.6K tok), XHIGH 2 (44.5K tok), HIGH 1 (40.2K tok), MEDIUM 2 (13.7K tok), NONE 2 (2.7K tok). Anthropic: 8 rows — HIGH 7 (41.3K tok), NONE 1 (16.0K tok).
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 2951.32
Task $ $2.86415
$ / rating point $0.0009705
#2 Performance: GPT-6.1 Sol (2743.75) · $0.596828/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / rating point $0.00001144
Score 1576.92
Task $ $0.018045
#2 Value: Claude Sonnet 5.5 ($0.00006098) · score 1819.12
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 2743.75
Task $ $0.596828
$ / rating point $0.0002175
#2 Elite: GPT-6 Sol (2462) · $0.843984/task
| # | ||||
|---|---|---|---|---|
| 1 raw #8 | 1576.92 | $0.018045 | $0.00001144 | |
| 2 raw #32 | 1819.12 | $0.110921 | $0.00006098 | |
| 3 raw #45 | 2147.38 | $0.271178 | $0.0001263 | |
| 4 raw #46 | 1688.35 | $0.220853 | $0.0001308 | |
| 5 raw #50 | 1508.08 | $0.24005 | $0.0001592 | |
| 6 raw #53 | 1520.72 | $0.26258 | $0.0001727 | |
| 7 raw #56 | 1589.38 | $0.328965 | $0.000207 | |
| 8 raw #60 | 1667.4 | $0.359171 | $0.0002154 | |
| 9 raw #61 | 2743.75 | $0.596828 | $0.0002175 | |
| 10 raw #73 | 1655.22 | $0.559265 | $0.0003379 | |
| 11 raw #75 | 2462 | $0.843984 | $0.0003428 | |
| 12 raw #79 | 1607 | $0.612174 | $0.0003809 | |
| 13 raw #83 | 1563.83 | $0.963945 | $0.0006164 | |
| 14 raw #84 | 1951.42 | $1.25405 | $0.0006426 | |
| 15 raw #87 | 2176.88 | $1.53408 | $0.0007047 | |
| 16 raw #88 | 1942.97 | $1.51713 | $0.0007808 | |
| 17 raw #89 | 1524.47 | $1.28155 | $0.0008407 | |
| 18 raw #90 | 2164.57 | $1.89051 | $0.0008734 | |
| 19 raw #91 | 2951.32 | $2.86415 | $0.0009705 | |
| 20 raw #92 | 2041.3 | $2.10882 | $0.001033 | |
| 21 raw #93 | 2143.2 | $2.29486 | $0.001071 | |
| 22 raw #1 | 566.05 | $0.000717 | $0.000001267 | |
| 23 raw #2 | 575.62 | $0.001303 | $0.000002264 | |
| 24 raw #3 | 324.98 | $0.000882 | $0.000002714 | |
| 25 raw #4 | 678.2 | $0.005179 | $0.000007636 | |
| 26 raw #5 | 172.97 | $0.001369 | $0.000007915 | |
| 27 raw #6 | 461.45 | $0.003677 | $0.000007968 | |
| 28 raw #7 | 718.67 | $0.007194 | $0.00001001 | |
| 29 raw #9 | 1306.08 | $0.019526 | $0.00001495 | |
| 30 raw #10 | 587.73 | $0.008947 | $0.00001522 | |
| 31 raw #11 | 264.7 | $0.005104 | $0.00001928 | |
| 32 raw #12 | 137.78 | $0.00278 | $0.00002018 | |
| 33 raw #13 | 394.93 | $0.009221 | $0.00002335 | |
| 34 raw #14 | 745.17 | $0.01883 | $0.00002527 | |
| 35 raw #15 | 799.77 | $0.020554 | $0.0000257 | |
| 36 raw #16 | 1004.52 | $0.027084 | $0.00002696 | |
| 37 raw #17 | 267.12 | $0.008051 | $0.00003014 | |
| 38 raw #18 | 456.5 | $0.013921 | $0.0000305 | |
| 39 raw #19 | 497.62 | $0.015587 | $0.00003132 | |
| 40 raw #20 | 797.73 | $0.025178 | $0.00003156 | |
| 41 raw #21 | 521.67 | $0.0166 | $0.00003182 | |
| 42 raw #22 | 210.18 | $0.007372 | $0.00003507 | |
| 43 raw #23 | 1086.03 | $0.052314 | $0.00004817 | |
| 44 raw #24 | 558.1 | $0.026917 | $0.00004823 | |
| 45 raw #25 | 1403.15 | $0.072764 | $0.00005186 | |
| 46 raw #26 | 826.17 | $0.044719 | $0.00005413 | |
| 47 raw #27 | 807.65 | $0.043744 | $0.00005416 | |
| 48 raw #28 | 804.12 | $0.043794 | $0.00005446 | |
| 49 raw #29 | 821.65 | $0.047458 | $0.00005776 | |
| 50 raw #30 | 944.17 | $0.055799 | $0.0000591 | |
| 51 raw #31 | 765.27 | $0.045726 | $0.00005975 | |
| 52 raw #33 | 996.5 | $0.061688 | $0.0000619 | |
| 53 raw #34 | 1127.58 | $0.073484 | $0.00006517 | |
| 54 raw #35 | 1367.2 | $0.091067 | $0.00006661 | |
| 55 raw #36 | 1092.35 | $0.081663 | $0.00007476 | |
| 56 raw #37 | 670.15 | $0.057069 | $0.00008516 | |
| 57 raw #38 | 661.88 | $0.068155 | $0.000103 | |
| 58 raw #39 | 326.4 | $0.035628 | $0.0001092 | |
| 59 raw #40 | 653.48 | $0.071831 | $0.0001099 | |
| 60 raw #41 | 933.55 | $0.103474 | $0.0001108 | |
| 61 raw #42 | 1249.83 | $0.142947 | $0.0001144 | |
| 62 raw #43 | 349.45 | $0.041945 | $0.00012 | |
| 63 raw #44 | 1150.28 | $0.13838 | $0.0001203 | |
| 64 raw #47 | 1309.28 | $0.176767 | $0.000135 | |
| 65 raw #48 | 597.5 | $0.087819 | $0.000147 | |
| 66 raw #49 | 1006.08 | $0.148062 | $0.0001472 | |
| 67 raw #51 | 1323.05 | $0.224549 | $0.0001697 | |
| 68 raw #52 | 221.8 | $0.03787 | $0.0001707 | |
| 69 raw #54 | 1092.67 | $0.196861 | $0.0001802 | |
| 70 raw #55 | 796.15 | $0.151558 | $0.0001904 | |
| 71 raw #57 | 886.23 | $0.184743 | $0.0002085 | |
| 72 raw #58 | 1208.83 | $0.253186 | $0.0002094 | |
| 73 raw #59 | 1162.45 | $0.244948 | $0.0002107 | |
| 74 raw #62 | 1176.75 | $0.259385 | $0.0002204 | |
| 75 raw #63 | 1188.58 | $0.271408 | $0.0002283 | |
| 76 raw #64 | 1293.55 | $0.296865 | $0.0002295 | |
| 77 raw #65 | 1192.15 | $0.315236 | $0.0002644 | |
| 78 raw #66 | 904.3 | $0.243628 | $0.0002694 | |
| 79 raw #67 | 1269.72 | $0.344275 | $0.0002711 | |
| 80 raw #68 | 655.35 | $0.185869 | $0.0002836 | |
| 81 raw #69 | 1025.38 | $0.292547 | $0.0002853 | |
| 82 raw #70 | 1411.83 | $0.421746 | $0.0002987 | |
| 83 raw #71 | 1189.4 | $0.3845 | $0.0003233 | |
| 84 raw #72 | 1160.6 | $0.388403 | $0.0003347 | |
| 85 raw #74 | 715.52 | $0.244468 | $0.0003417 | |
| 86 raw #76 | 785.52 | $0.26988 | $0.0003436 | |
| 87 raw #77 | 763.98 | $0.272986 | $0.0003573 | |
| 88 raw #78 | 1213.15 | $0.456946 | $0.0003767 | |
| 89 raw #80 | 1244.92 | $0.512093 | $0.0004113 | |
| 90 raw #81 | 1299.9 | $0.587753 | $0.0004522 | |
| 91 raw #82 | 911.02 | $0.441921 | $0.0004851 | |
| 92 raw #85 | 674.77 | $0.434347 | $0.0006437 | |
| 93 raw #86 | 1463.12 | $0.96046 | $0.0006564 | |
| 94 raw #94 | 1327.3 | $1.77075 | $0.001334 |
task cost divided by ALE-Bench rating — dollars per rating point. 94 ranked, 1 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score
REASONING · PROOFBENCH ℹ️ EFFORT NOT STATED FOR GOOGLE DEEPMIND
Formal Mathematical Proof & Deep Reasoning — ProofBench
Metric: accuracy (Epoch AI) · Unit cost ratio: $ / correct answer · Cost basis: unknown (leaderboard-maintainer-verified).
ℹ️ Methodological Effort Note: effort not stated upstream
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). The measurer did not state the effort for most of one lab's rows, so effort parity cannot be checked. Anthropic: 8 rows — MAX 8. OpenAI: 7 rows — MAX 1, XHIGH 3, UNSTATED 3. Google DeepMind: 3 rows — UNSTATED 3.
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 1
Task $ $0.558327
$ / correct answer $0.5583
#2 Performance: Claude Opus 5.5 (1) · $0.961648/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / correct answer $0.06506
Score 0.64
Task $ $0.041637
#2 Value: DeepSeek-V4 Flash (0731) ($0.0653) · score 0.56
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 0.89
Task $ $0.173874
$ / correct answer $0.1954
#2 Elite: GPT-6.1 Sol (0.99) · $0.233057/task
| # | ||||
|---|---|---|---|---|
| 1 raw #1 | 0.64 | $0.041637 | $0.06506 | |
| 2 raw #2 | 0.56 | $0.036568 | $0.0653 | |
| 3 raw #3 | 0.5 | $0.062191 | $0.1244 | |
| 4 raw #4 | 0.89 | $0.173874 | $0.1954 | |
| 5 raw #5 | 0.6 | $0.123288 | $0.2055 | |
| 6 raw #7 | 0.54 | $0.126437 | $0.2341 | |
| 7 raw #8 | 0.99 | $0.233057 | $0.2354 | |
| 8 raw #12 | 1 | $0.558327 | $0.5583 | |
| 9 raw #13 | 0.83 | $0.476321 | $0.5739 | |
| 10 raw #14 | Muse Spark 1.3 CAPABLE | 0.55 | $0.433712 | $0.7886 |
| 11 raw #15 | 0.99 | $0.782978 | $0.7909 | |
| 12 raw #16 | 0.58 | $0.479019 | $0.8259 | |
| 13 raw #17 | 1 | $0.961648 | $0.9616 | |
| 14 raw #18 | 0.58 | $0.558508 | $0.9629 | |
| 15 raw #26 | 0.51 | $0.759043 | $1.488 | |
| 16 raw #28 | Muse Wonder 1.0 ELITE | 0.84 | $1.31334 | $1.564 |
| 17 raw #29 | 0.74 | $1.27421 | $1.722 | |
| 18 raw #30 | 0.99 | $1.71238 | $1.73 | |
| 19 raw #31 | 0.83 | $1.43831 | $1.733 | |
| 20 raw #32 | 0.77 | $1.37368 | $1.784 | |
| 21 raw #36 | 0.87 | $1.6468 | $1.893 | |
| 22 raw #37 | 0.99 | $2.10619 | $2.127 | |
| 23 raw #43 | 0.58 | $1.50398 | $2.593 | |
| 24 raw #44 | 1 | $2.70347 | $2.703 | |
| 25 raw #46 | 0.69 | $2.29313 | $3.323 | |
| 26 raw #47 | 0.54 | $1.80779 | $3.348 | |
| 27 raw #48 | 0.5 | $1.68179 | $3.364 | |
| 28 raw #53 | 0.5 | $2.70358 | $5.407 | |
| 29 raw #54 | 0.56 | $3.20201 | $5.718 | |
| 30 raw #55 | 0.95 | $5.45152 | $5.738 | |
| 31 raw #6 | 0.16 | $0.033594 | $0.21 | |
| 32 raw #9 | 0.05 | $0.015894 | $0.3179 | |
| 33 raw #10 | Muse Spark LOW SCORE | 0.17 | $0.067086 | $0.3946 |
| 34 raw #11 | 0.12 | $0.049218 | $0.4102 | |
| 35 raw #19 | 0.13 | $0.132788 | $1.021 | |
| 36 raw #20 | Muse Spark 1.2 LOW SCORE | 0.43 | $0.442007 | $1.028 |
| 37 raw #21 | 0.21 | $0.249441 | $1.188 | |
| 38 raw #22 | 0.48 | $0.602872 | $1.256 | |
| 39 raw #23 | 0.39 | $0.519718 | $1.333 | |
| 40 raw #24 | 0.36 | $0.522059 | $1.45 | |
| 41 raw #25 | 0.08 | $0.11773 | $1.472 | |
| 42 raw #27 | 0.34 | $0.518303 | $1.524 | |
| 43 raw #33 | 0.31 | $0.569863 | $1.838 | |
| 44 raw #34 | 0.16 | $0.295811 | $1.849 | |
| 45 raw #35 | 0.31 | $0.573991 | $1.852 | |
| 46 raw #38 | 0.15 | $0.327286 | $2.182 | |
| 47 raw #39 | 0.11 | $0.244122 | $2.219 | |
| 48 raw #40 | 0.14 | $0.31815 | $2.272 | |
| 49 raw #41 | 0.04 | $0.096578 | $2.414 | |
| 50 raw #42 | 0.09 | $0.22021 | $2.447 | |
| 51 raw #45 | 0.26 | $0.784253 | $3.016 | |
| 52 raw #49 | 0.19 | $0.65341 | $3.439 | |
| 53 raw #50 | 0.36 | $1.26704 | $3.52 | |
| 54 raw #51 | 0.18 | $0.852361 | $4.735 | |
| 55 raw #52 | 0.45 | $2.16211 | $4.805 | |
| 56 raw #56 | 0.09 | $0.527356 | $5.86 | |
| 57 raw #57 | 0.2 | $1.22908 | $6.145 | |
| 58 raw #58 | 0.1 | $0.642365 | $6.424 | |
| 59 raw #59 | 0.15 | $1.05969 | $7.065 | |
| 60 raw #60 | 0.26 | $2.46207 | $9.469 | |
| 61 raw #61 | 0.16 | $2.29883 | $14.37 | |
| 62 raw #62 | 0.09 | $1.30203 | $14.47 |
task cost divided by accuracy — expected spend per correct answer at this accuracy. 62 ranked, 0 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score
AGENT · DEEPSWE ✅ EFFORT PARITY (HIGH · MEDIUM · LOW)
Autonomous SWE Agent & Repository Resolution — DeepSWE
Metric: pass@1 (Epoch AI) · Unit cost ratio: $ / solved task · Cost basis: unknown (leaderboard-maintainer-verified).
⚖️ Effort Parity: frontier labs measured at shared effort tiers
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). Every compared lab's rows sit at an effort another lab was also measured at. Anthropic: 20 rows — MAX 4 (126.8K tok), XHIGH 4 (88.9K tok), HIGH 4 (60.7K tok), MEDIUM 4 (40.8K tok), LOW 4 (27.1K tok). OpenAI: 20 rows — MAX 3 (61.1K tok), XHIGH 5 (40.7K tok), HIGH 4 (27.5K tok), MEDIUM 4 (19.0K tok), LOW 4 (10.0K tok). Google DeepMind: 5 rows — HIGH 2 (125.2K tok), MEDIUM 2 (109.3K tok), LOW 1 (73.4K tok).
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 0.741
Task $ $4.42912
$ / solved task $5.976
#2 Performance: Gemini 3.8 Flash (0.738) · $2.36235/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / solved task $1.758
Score 0.442
Task $ $0.777901
#2 Value: GPT-5.6 Terra ($2.11) · score 0.538
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 0.67
Task $ $1.59522
$ / solved task $2.38
#2 Elite: GPT-5.6 Luna (0.569) · $1.53562/task
| # | ||||
|---|---|---|---|---|
| 1 raw #2 | 0.442 | $0.777901 | $1.758 | |
| 2 raw #5 | 0.538 | $1.13438 | $2.11 | |
| 3 raw #6 | 0.454 | $1.07427 | $2.369 | |
| 4 raw #7 | 0.67 | $1.59522 | $2.38 | |
| 5 raw #8 | 0.416 | $1.04236 | $2.503 | |
| 6 raw #9 | 0.569 | $1.53562 | $2.701 | |
| 7 raw #10 | 0.71 | $1.9671 | $2.77 | |
| 8 raw #11 | 0.581 | $1.66262 | $2.86 | |
| 9 raw #12 | 0.611 | $1.86203 | $3.049 | |
| 10 raw #13 | 0.655 | $2.02514 | $3.092 | |
| 11 raw #14 | 0.738 | $2.36235 | $3.2 | |
| 12 raw #15 | 0.653 | $2.17629 | $3.335 | |
| 13 raw #16 | 0.538 | $1.83226 | $3.408 | |
| 14 raw #17 | 0.602 | $2.12719 | $3.535 | |
| 15 raw #18 | 0.728 | $3.07547 | $4.225 | |
| 16 raw #19 | 0.533 | $2.36114 | $4.428 | |
| 17 raw #21 | 0.538 | $2.41575 | $4.493 | |
| 18 raw #22 | 0.672 | $3.02812 | $4.507 | |
| 19 raw #24 | 0.689 | $3.28979 | $4.774 | |
| 20 raw #25 | 0.694 | $3.46983 | $5 | |
| 21 raw #26 | 0.54 | $2.74934 | $5.093 | |
| 22 raw #27 | 0.675 | $3.44898 | $5.111 | |
| 23 raw #28 | 0.732 | $3.92372 | $5.358 | |
| 24 raw #29 | 0.408 | $2.29337 | $5.621 | |
| 25 raw #30 | 0.741 | $4.42912 | $5.976 | |
| 26 raw #31 | 0.596 | $3.75787 | $6.307 | |
| 27 raw #32 | 0.575 | $3.72906 | $6.49 | |
| 28 raw #33 | 0.707 | $4.70366 | $6.65 | |
| 29 raw #34 | 0.652 | $4.38487 | $6.726 | |
| 30 raw #35 | 0.549 | $3.6955 | $6.735 | |
| 31 raw #36 | 0.685 | $4.65468 | $6.794 | |
| 32 raw #37 | 0.487 | $3.44386 | $7.076 | |
| 33 raw #38 | 0.696 | $4.94585 | $7.104 | |
| 34 raw #40 | 0.644 | $5.10044 | $7.922 | |
| 35 raw #41 | 0.667 | $5.49767 | $8.237 | |
| 36 raw #42 | 0.518 | $4.28194 | $8.271 | |
| 37 raw #43 | 0.728 | $6.07608 | $8.343 | |
| 38 raw #45 | 0.654 | $6.08819 | $9.314 | |
| 39 raw #46 | 0.467 | $4.41901 | $9.466 | |
| 40 raw #48 | 0.732 | $7.49783 | $10.24 | |
| 41 raw #49 | 0.398 | $4.07904 | $10.25 | |
| 42 raw #50 | 0.67 | $7.22624 | $10.78 | |
| 43 raw #51 | 0.518 | $5.65246 | $10.92 | |
| 44 raw #52 | 0.727 | $8.38644 | $11.54 | |
| 45 raw #53 | 0.732 | $9.0722 | $12.4 | |
| 46 raw #54 | 0.686 | $9.17764 | $13.38 | |
| 47 raw #55 | 0.544 | $8.00636 | $14.73 | |
| 48 raw #56 | 0.482 | $7.42556 | $15.4 | |
| 49 raw #57 | 0.736 | $11.8376 | $16.07 | |
| 50 raw #60 | 0.699 | $13.4145 | $19.19 | |
| 51 raw #61 | 0.374 | $7.34182 | $19.64 | |
| 52 raw #62 | 0.59 | $13.2226 | $22.42 | |
| 53 raw #63 | 0.497 | $11.8906 | $23.94 | |
| 54 raw #64 | 0.697 | $21.6347 | $31.03 | |
| 55 raw #65 | 0.538 | $26.3999 | $49.03 | |
| 56 raw #1 | 0.351 | $0.583249 | $1.661 | |
| 57 raw #3 | 0.241 | $0.427745 | $1.778 | |
| 58 raw #4 | 0.113 | $0.21631 | $1.917 | |
| 59 raw #20 | 0.27 | $1.2002 | $4.447 | |
| 60 raw #23 | 0.0155 | $0.0724062 | $4.675 | |
| 61 raw #39 | 0.305 | $2.18656 | $7.166 | |
| 62 raw #44 | 0.305 | $2.81554 | $9.222 | |
| 63 raw #47 | 0.361 | $3.44671 | $9.558 | |
| 64 raw #58 | 0.117 | $2.1434 | $18.28 | |
| 65 raw #59 | 0.299 | $5.52235 | $18.45 |
task cost divided by pass@1 — expected spend per solved task at this pass rate. 65 ranked, 0 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score
IDE · CURSORBENCH ✅ EFFORT PARITY (HIGH · MEDIUM · LOW)
IDE Code Editing & Multi-Step Completion — CursorBench
Metric: accuracy (Epoch AI) · Unit cost ratio: $ / correct answer · Cost basis: unknown (leaderboard-maintainer-verified).
⚖️ Effort Parity: frontier labs measured at shared effort tiers
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). Every compared lab's rows sit at an effort another lab was also measured at. Anthropic: 30 rows — MAX 6, XHIGH 6, HIGH 6, MEDIUM 6, LOW 6. OpenAI: 10 rows — MAX 2, XHIGH 2, HIGH 2, MEDIUM 2, LOW 2. Google DeepMind: 6 rows — HIGH 2, MEDIUM 2, LOW 1, MINIMAL 1. Meta AI: 6 rows — MAX 1, XHIGH 1, HIGH 1, MEDIUM 1, LOW 1, MINIMAL 1.
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 0.578
Task $ $13.43
$ / correct answer $23.24
#2 Performance: Claude Sonnet 5.5 (0.555) · $7.05/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / correct answer $0.2589
Score 0.309
Task $ $0.08
#2 Value: GPT-5.6 Luna ($0.8503) · score 0.294
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 0.443
Task $ $0.56
$ / correct answer $1.264
#2 Elite: Claude Sonnet 5.5 (0.478) · $1.2/task
| # | ||||
|---|---|---|---|---|
| 1 raw #2 | 0.309 | $0.08 | $0.2589 | |
| 2 raw #4 | 0.369 | $0.17 | $0.4607 | |
| 3 raw #5 | 0.423 | $0.32 | $0.7565 | |
| 4 raw #6 | 0.294 | $0.25 | $0.8503 | |
| 5 raw #7 | 0.358 | $0.37 | $1.034 | |
| 6 raw #8 | 0.443 | $0.56 | $1.264 | |
| 7 raw #9 | 0.392 | $0.52 | $1.327 | |
| 8 raw #10 | 0.33 | $0.44 | $1.333 | |
| 9 raw #13 | 0.484 | $1.12 | $2.314 | |
| 10 raw #15 | 0.478 | $1.2 | $2.51 | |
| 11 raw #16 | 0.437 | $1.17 | $2.677 | |
| 12 raw #17 | 0.359 | $1.03 | $2.869 | |
| 13 raw #18 | 0.293 | $0.93 | $3.174 | |
| 14 raw #20 | 0.307 | $1.11 | $3.616 | |
| 15 raw #21 | 0.326 | $1.49 | $4.571 | |
| 16 raw #22 | 0.331 | $1.58 | $4.773 | |
| 17 raw #23 | 0.334 | $1.66 | $4.97 | |
| 18 raw #24 | 0.531 | $2.81 | $5.292 | |
| 19 raw #25 | 0.336 | $1.81 | $5.387 | |
| 20 raw #26 | 0.525 | $2.91 | $5.543 | |
| 21 raw #27 | 0.375 | $2.1 | $5.6 | |
| 22 raw #28 | 0.311 | $1.77 | $5.691 | |
| 23 raw #30 | 0.416 | $2.64 | $6.346 | |
| 24 raw #31 | 0.334 | $2.25 | $6.737 | |
| 25 raw #32 | 0.56 | $3.97 | $7.089 | |
| 26 raw #33 | 0.488 | $3.54 | $7.254 | |
| 27 raw #34 | 0.357 | $2.85 | $7.983 | |
| 28 raw #35 | 0.506 | $4.09 | $8.083 | |
| 29 raw #37 | 0.528 | $4.39 | $8.314 | |
| 30 raw #38 | 0.416 | $3.49 | $8.389 | |
| 31 raw #39 | 0.529 | $4.57 | $8.639 | |
| 32 raw #40 | 0.361 | $3.48 | $9.64 | |
| 33 raw #41 | 0.439 | $4.69 | $10.68 | |
| 34 raw #42 | 0.373 | $4.06 | $10.88 | |
| 35 raw #43 | 0.308 | $3.48 | $11.3 | |
| 36 raw #44 | 0.377 | $4.4 | $11.67 | |
| 37 raw #45 | 0.396 | $4.7 | $11.87 | |
| 38 raw #46 | 0.407 | $4.87 | $11.97 | |
| 39 raw #47 | 0.451 | $5.44 | $12.06 | |
| 40 raw #48 | 0.413 | $5.14 | $12.45 | |
| 41 raw #49 | 0.56 | $6.98 | $12.46 | |
| 42 raw #50 | 0.555 | $7.05 | $12.7 | |
| 43 raw #51 | 0.404 | $5.2 | $12.87 | |
| 44 raw #52 | 0.463 | $6.01 | $12.98 | |
| 45 raw #53 | 0.32 | $4.55 | $14.22 | |
| 46 raw #54 | 0.414 | $6.1 | $14.73 | |
| 47 raw #55 | 0.468 | $7.05 | $15.06 | |
| 48 raw #56 | 0.433 | $6.94 | $16.03 | |
| 49 raw #57 | 0.492 | $9.08 | $18.46 | |
| 50 raw #58 | 0.417 | $8.23 | $19.74 | |
| 51 raw #59 | 0.447 | $9 | $20.13 | |
| 52 raw #60 | 0.341 | $7.17 | $21.03 | |
| 53 raw #61 | 0.578 | $13.43 | $23.24 | |
| 54 raw #62 | 0.461 | $11.43 | $24.79 | |
| 55 raw #63 | 0.516 | $13.01 | $25.21 | |
| 56 raw #64 | 0.466 | $11.95 | $25.64 | |
| 57 raw #65 | 0.518 | $17.28 | $33.36 | |
| 58 raw #1 | 0.16 | $0.03 | $0.1875 | |
| 59 raw #3 | 0.222 | $0.08 | $0.3604 | |
| 60 raw #11 | 0.252 | $0.52 | $2.063 | |
| 61 raw #12 | 0.243 | $0.56 | $2.305 | |
| 62 raw #14 | 0.276 | $0.64 | $2.319 | |
| 63 raw #19 | 0.246 | $0.87 | $3.537 | |
| 64 raw #29 | 0.241 | $1.39 | $5.768 | |
| 65 raw #36 | 0.28 | $2.31 | $8.25 |
task cost divided by accuracy — expected spend per correct answer at this accuracy. 65 ranked, 0 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score
ML-CODE · WEIRDML ✅ EFFORT PARITY (MAX · XHIGH · HIGH · NONE)
Algorithmic ML & Scientific Coding — WeirdML
Metric: accuracy (Epoch AI) · Unit cost ratio: $ / correct answer · Cost basis: unknown (leaderboard-maintainer-verified).
⚖️ Effort Parity: frontier labs measured at shared effort tiers
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). At least 82% of every compared lab's rows sit at an effort another lab was also measured at. OpenAI: 15 rows — MAX 4, XHIGH 3, HIGH 4, NONE 3, UNSTATED 1. Anthropic: 11 rows — MAX 3, XHIGH 1, HIGH 4, MEDIUM 1, NONE 1, UNSTATED 1.
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 0.936
Task $ $29.3058
$ / correct answer $31.32
#2 Performance: Claude Fable 5.1 (0.929) · $11.6866/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / correct answer $0.04442
Score 0.571
Task $ $0.0253424
#2 Value: GPT-5.4 nano ($0.09699) · score 0.492
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 0.783
Task $ $0.311906
$ / correct answer $0.3985
#2 Elite: Gemini 3.8 Flash (0.848) · $0.507268/task
| # | ||||
|---|---|---|---|---|
| 1 raw #18 | 0.571 | $0.0253424 | $0.04442 | |
| 2 raw #20 | 0.63 | $0.0285783 | $0.04539 | |
| 3 raw #34 | 0.492 | $0.0477472 | $0.09699 | |
| 4 raw #40 | 0.522 | $0.066424 | $0.1273 | |
| 5 raw #47 | 0.499 | $0.0903199 | $0.181 | |
| 6 raw #52 | 0.527 | $0.109767 | $0.2084 | |
| 7 raw #54 | 0.523 | $0.126283 | $0.2416 | |
| 8 raw #60 | 0.662 | $0.191032 | $0.2885 | |
| 9 raw #62 | 0.609 | $0.19999 | $0.3286 | |
| 10 raw #64 | 0.616 | $0.221588 | $0.3597 | |
| 11 raw #65 | 0.526 | $0.195685 | $0.3723 | |
| 12 raw #66 | 0.783 | $0.311906 | $0.3985 | |
| 13 raw #68 | 0.541 | $0.231634 | $0.428 | |
| 14 raw #69 | 0.524 | $0.226067 | $0.4313 | |
| 15 raw #70 | 0.496 | $0.214626 | $0.4327 | |
| 16 raw #71 | 0.496 | $0.215 | $0.4335 | |
| 17 raw #72 | 0.574 | $0.267909 | $0.4664 | |
| 18 raw #73 | 0.574 | $0.267909 | $0.4664 | |
| 19 raw #74 | 0.673 | $0.334832 | $0.4976 | |
| 20 raw #76 | 0.848 | $0.507268 | $0.5981 | |
| 21 raw #79 | 0.559 | $0.350354 | $0.6272 | |
| 22 raw #80 | 0.603 | $0.380109 | $0.6304 | |
| 23 raw #82 | 0.634 | $0.415976 | $0.6557 | |
| 24 raw #84 | 0.705 | $0.4752 | $0.6745 | |
| 25 raw #85 | 0.489 | $0.3403 | $0.6959 | |
| 26 raw #86 | 0.477 | $0.334378 | $0.7009 | |
| 27 raw #89 | 0.699 | $0.525771 | $0.7519 | |
| 28 raw #90 | 0.54 | $0.40966 | $0.7582 | |
| 29 raw #92 | 0.672 | $0.538951 | $0.8026 | |
| 30 raw #93 | 0.764 | $0.6895 | $0.9025 | |
| 31 raw #94 | 0.764 | $0.689518 | $0.9025 | |
| 32 raw #95 | 0.607 | $0.59318 | $0.9772 | |
| 33 raw #97 | 0.561 | $0.622712 | $1.11 | |
| 34 raw #98 | 0.76 | $0.847294 | $1.114 | |
| 35 raw #99 | 0.659 | $0.735544 | $1.117 | |
| 36 raw #100 | 0.608 | $0.691335 | $1.138 | |
| 37 raw #101 | 0.637 | $0.738304 | $1.158 | |
| 38 raw #102 | 0.626 | $0.756041 | $1.207 | |
| 39 raw #104 | 0.688 | $0.895463 | $1.302 | |
| 40 raw #105 | 0.888 | $1.2548 | $1.414 | |
| 41 raw #106 | 0.863 | $1.41957 | $1.644 | |
| 42 raw #109 | 0.721 | $1.36019 | $1.887 | |
| 43 raw #110 | 0.545 | $1.09561 | $2.009 | |
| 44 raw #111 | 0.545 | $1.09561 | $2.009 | |
| 45 raw #112 | 0.603 | $1.29102 | $2.141 | |
| 46 raw #113 | 0.764 | $1.82071 | $2.382 | |
| 47 raw #114 | 0.916 | $2.19335 | $2.395 | |
| 48 raw #115 | 0.826 | $2.05366 | $2.487 | |
| 49 raw #116 | 0.839 | $2.22158 | $2.648 | |
| 50 raw #117 | 0.87 | $2.31221 | $2.659 | |
| 51 raw #118 | 0.829 | $2.34981 | $2.835 | |
| 52 raw #119 | 0.722 | $2.04738 | $2.836 | |
| 53 raw #120 | 0.793 | $2.35 | $2.963 | |
| 54 raw #121 | 0.849 | $2.56835 | $3.025 | |
| 55 raw #122 | 0.779 | $2.45258 | $3.148 | |
| 56 raw #123 | 0.929 | $2.97541 | $3.204 | |
| 57 raw #124 | 0.923 | $3.20134 | $3.468 | |
| 58 raw #125 | 0.878 | $3.1069 | $3.537 | |
| 59 raw #127 | 0.476 | $1.84876 | $3.887 | |
| 60 raw #130 | 0.661 | $2.85935 | $4.328 | |
| 61 raw #131 | 0.894 | $3.87074 | $4.328 | |
| 62 raw #132 | 0.582 | $2.54038 | $4.364 | |
| 63 raw #136 | 0.918 | $4.74195 | $5.167 | |
| 64 raw #137 | 0.777 | $5.71994 | $7.362 | |
| 65 raw #138 | 0.78 | $5.77937 | $7.414 | |
| 66 raw #139 | 0.779 | $5.78 | $7.42 | |
| 67 raw #140 | 0.933 | $7.74851 | $8.307 | |
| 68 raw #141 | 0.919 | $8.42441 | $9.163 | |
| 69 raw #142 | 0.755 | $7.95262 | $10.54 | |
| 70 raw #143 | 0.929 | $11.6866 | $12.58 | |
| 71 raw #144 | 0.604 | $7.64764 | $12.66 | |
| 72 raw #145 | 0.936 | $29.3058 | $31.32 | |
| 73 raw #1 | 0.144 | $0.00146949 | $0.01018 | |
| 74 raw #2 | 0.368 | $0.00380622 | $0.01033 | |
| 75 raw #3 | 0.19 | $0.00258172 | $0.0136 | |
| 76 raw #4 | 0.438 | $0.00727954 | $0.01664 | |
| 77 raw #5 | 0.259 | $0.00447468 | $0.01728 | |
| 78 raw #6 | 0.0173 | $0.000410396 | $0.02372 | |
| 79 raw #7 | Llama 4 Maverick LOW SCORE | 0.245 | $0.00613222 | $0.02506 |
| 80 raw #8 | 0.258 | $0.00673442 | $0.02613 | |
| 81 raw #9 | 0.409 | $0.0108003 | $0.02639 | |
| 82 raw #10 | 0.297 | $0.00814814 | $0.02739 | |
| 83 raw #11 | 0.419 | $0.011485 | $0.02742 | |
| 84 raw #12 | 0.429 | $0.0127262 | $0.02969 | |
| 85 raw #13 | 0.456 | $0.014616 | $0.03203 | |
| 86 raw #14 | 0.387 | $0.0134071 | $0.03464 | |
| 87 raw #15 | Llama 3.1 70B Instruct LOW SCORE | 0.0897 | $0.00322267 | $0.03593 |
| 88 raw #16 | 0.361 | $0.013904 | $0.03854 | |
| 89 raw #17 | 0.376 | $0.0166067 | $0.04415 | |
| 90 raw #19 | 0.118 | $0.00526076 | $0.04473 | |
| 91 raw #21 | 0.38 | $0.0199518 | $0.05256 | |
| 92 raw #22 | 0.375 | $0.0197308 | $0.05262 | |
| 93 raw #23 | 0.373 | $0.0208753 | $0.056 | |
| 94 raw #24 | 0.351 | $0.0234918 | $0.067 | |
| 95 raw #25 | 0.381 | $0.0262203 | $0.06889 | |
| 96 raw #26 | 0.16 | $0.0113277 | $0.07093 | |
| 97 raw #27 | 0.467 | $0.0352106 | $0.07535 | |
| 98 raw #28 | 0.384 | $0.0311958 | $0.08132 | |
| 99 raw #29 | 0.352 | $0.0302532 | $0.0859 | |
| 100 raw #30 | 0.394 | $0.0342997 | $0.08714 | |
| 101 raw #31 | 0.113 | $0.010191 | $0.09019 | |
| 102 raw #32 | Llama 3.1 405B Instruct LOW SCORE | 0.214 | $0.0202977 | $0.09494 |
| 103 raw #33 | 0.344 | $0.0328304 | $0.09544 | |
| 104 raw #35 | 0.395 | $0.040395 | $0.1022 | |
| 105 raw #36 | 0.367 | $0.0418198 | $0.114 | |
| 106 raw #37 | 0.456 | $0.0526125 | $0.1154 | |
| 107 raw #38 | 0.39 | $0.0467557 | $0.1199 | |
| 108 raw #39 | 0.331 | $0.0401798 | $0.1213 | |
| 109 raw #41 | 0.0984 | $0.0140751 | $0.143 | |
| 110 raw #42 | 0.416 | $0.0710683 | $0.1707 | |
| 111 raw #43 | 0.345 | $0.0596161 | $0.1729 | |
| 112 raw #44 | 0.464 | $0.0804629 | $0.1733 | |
| 113 raw #45 | 0.419 | $0.073245 | $0.1748 | |
| 114 raw #46 | 0.412 | $0.0725924 | $0.1763 | |
| 115 raw #48 | 0.307 | $0.0574802 | $0.187 | |
| 116 raw #49 | 0.428 | $0.0809377 | $0.1892 | |
| 117 raw #50 | 0.379 | $0.0728587 | $0.1922 | |
| 118 raw #51 | 0.454 | $0.0893874 | $0.1969 | |
| 119 raw #53 | 0.365 | $0.0785055 | $0.2151 | |
| 120 raw #55 | 0.39 | $0.0970296 | $0.2485 | |
| 121 raw #56 | 0.41 | $0.105854 | $0.2585 | |
| 122 raw #57 | 0.409 | $0.105854 | $0.2585 | |
| 123 raw #58 | 0.442 | $0.122786 | $0.2781 | |
| 124 raw #59 | 0.465 | $0.129944 | $0.2793 | |
| 125 raw #61 | 0.363 | $0.111275 | $0.3064 | |
| 126 raw #63 | 0.251 | $0.0835561 | $0.3326 | |
| 127 raw #67 | 0.437 | $0.174467 | $0.3992 | |
| 128 raw #75 | 0.222 | $0.114127 | $0.5132 | |
| 129 raw #77 | 0.467 | $0.279683 | $0.5989 | |
| 130 raw #78 | 0.4 | $0.248324 | $0.6213 | |
| 131 raw #81 | 0.437 | $0.278465 | $0.6371 | |
| 132 raw #83 | 0.372 | $0.247725 | $0.6652 | |
| 133 raw #87 | 0.439 | $0.308056 | $0.702 | |
| 134 raw #88 | 0.461 | $0.335986 | $0.7287 | |
| 135 raw #91 | 0.31 | $0.245601 | $0.793 | |
| 136 raw #96 | 0.457 | $0.451773 | $0.9879 | |
| 137 raw #103 | 0.0348 | $0.0423708 | $1.218 | |
| 138 raw #107 | 0.102 | $0.178291 | $1.755 | |
| 139 raw #108 | 0.18 | $0.321028 | $1.782 | |
| 140 raw #126 | 0.459 | $1.75555 | $3.828 | |
| 141 raw #128 | 0.437 | $1.73527 | $3.969 | |
| 142 raw #129 | 0.394 | $1.575 | $4.001 | |
| 143 raw #133 | 0.461 | $2.02867 | $4.402 | |
| 144 raw #134 | 0.192 | $0.856788 | $4.458 | |
| 145 raw #135 | 0.124 | $0.637622 | $5.159 |
task cost divided by accuracy — expected spend per correct answer at this accuracy. 145 ranked, 4 retired excluded, 0 dropped for missing cost, 0 for a published cost of $0, 0 for a non-positive score
ABSTRACT · ARC-AGI-2 ⚠️ ASYMMETRIC EFFORT (OPENAI XHIGH · ANTHROPIC HIGH · MOONSHOT MAX)
Abstract Fluid Intelligence & Pattern Discovery — ARC-AGI-2
Metric: accuracy (Epoch AI) · Unit cost ratio: $ / correct answer · Cost basis: unknown (community-submitted).
⚖️ Methodological Effort Alert: frontier labs measured at different effort tiers
Frontier-watch models on this tab, rows per reasoning effort (median output tokens per task where reported). Only 71% of one compared lab's rows sit at an effort another lab was also measured at; filter to one effort tier before comparing labs. OpenAI: 21 rows — MAX 2, XHIGH 6, HIGH 5, MEDIUM 4, LOW 4. Anthropic: 3 rows — HIGH 1, MEDIUM 1, LOW 1. Moonshot: 3 rows — MAX 1, HIGH 1, LOW 1.
🏆 PERFORMANCE LEADER (SOTA) Highest Raw Score
Score 0.925
Task $ $1.44
$ / correct answer $1.557
#2 Performance: GPT-5.5 (0.85) · $1.87/task
⚡ VALUE / COST LEADER Best Ratio at ≥50% Ceiling
$ / correct answer $0.7007
Score 0.671
Task $ $0.47
#2 Value: GPT-5.6 Terra ($0.8199) · score 0.671
💎 ELITE PARETO SWEET-SPOT Score ≥75% Ceiling · Lowest Cost
Score 0.854
Task $ $0.74
$ / correct answer $0.8663
#2 Elite: GPT-5.6 Terra (0.742) · $0.69/task
| # | ||||
|---|---|---|---|---|
| 1 raw #5 | 0.671 | $0.47 | $0.7007 | |
| 2 raw #8 | 0.671 | $0.55 | $0.8199 | |
| 3 raw #9 | 0.854 | $0.74 | $0.8663 | |
| 4 raw #12 | 0.742 | $0.69 | $0.9303 | |
| 5 raw #15 | 0.595 | $0.67 | $1.125 | |
| 6 raw #17 | 0.476 | $0.54 | $1.134 | |
| 7 raw #19 | 0.9 | $1.04 | $1.156 | |
| 8 raw #20 | 0.721 | $0.8503 | $1.18 | |
| 9 raw #23 | 0.704 | $0.86 | $1.221 | |
| 10 raw #24 | 0.554 | $0.68 | $1.227 | |
| 11 raw #25 | 0.839 | $1.09 | $1.299 | |
| 12 raw #28 | 0.651 | $0.92 | $1.412 | |
| 13 raw #29 | 0.526 | $0.75 | $1.425 | |
| 14 raw #30 | 0.526 | $0.78 | $1.482 | |
| 15 raw #32 | 0.675 | $1.02 | $1.511 | |
| 16 raw #33 | 0.925 | $1.44 | $1.557 | |
| 17 raw #36 | 0.55 | $0.947223 | $1.722 | |
| 18 raw #37 | 0.833 | $1.45 | $1.74 | |
| 19 raw #38 | 0.74 | $1.52 | $2.055 | |
| 20 raw #40 | 0.85 | $1.87 | $2.2 | |
| 21 raw #44 | 0.604 | $1.59269 | $2.636 | |
| 22 raw #45 | 0.622 | $1.6818 | $2.703 | |
| 23 raw #55 | 0.717 | $2.3896 | $3.334 | |
| 24 raw #62 | 0.529 | $1.8982 | $3.588 | |
| 25 raw #65 | 0.721 | $2.7425 | $3.805 | |
| 26 raw #66 | 0.621 | $2.38 | $3.834 | |
| 27 raw #69 | 0.675 | $2.96 | $4.385 | |
| 28 raw #71 | 0.604 | $2.6966 | $4.463 | |
| 29 raw #72 | 0.583 | $2.7179 | $4.66 | |
| 30 raw #94 | 0.846 | $10.51 | $12.43 | |
| 31 raw #98 | 0.842 | $10.76 | $12.79 | |
| 32 raw #110 | 0.833 | $16.41 | $19.69 | |
| 33 raw #115 | 0.542 | $15.721 | $29.03 | |
| 34 raw #1 | 0.013 | $0.004 | $0.3077 | |
| 35 raw #2 | 0.0125 | $0.0044 | $0.352 | |
| 36 raw #3 | 0.0167 | $0.0094 | $0.5629 | |
| 37 raw #4 | 0.0153 | $0.01 | $0.6536 | |
| 38 raw #6 | 0.375 | $0.27 | $0.72 | |
| 39 raw #7 | 0.425 | $0.32 | $0.7529 | |
| 40 raw #10 | 0.188 | $0.17 | $0.9067 | |
| 41 raw #11 | 0.292 | $0.27 | $0.9256 | |
| 42 raw #13 | 0.293 | $0.29 | $0.9894 | |
| 43 raw #14 | 0.333 | $0.35 | $1.05 | |
| 44 raw #16 | 0.0261 | $0.0295 | $1.13 | |
| 45 raw #18 | 0.0528 | $0.061 | $1.155 | |
| 46 raw #21 | 0.0889 | $0.1074 | $1.208 | |
| 47 raw #22 | 0.331 | $0.4 | $1.21 | |
| 48 raw #26 | 0.0514 | $0.07 | $1.362 | |
| 49 raw #27 | 0.0216 | $0.0302 | $1.398 | |
| 50 raw #31 | 0.0736 | $0.11 | $1.495 | |
| 51 raw #34 | 0.0088 | $0.0137 | $1.557 | |
| 52 raw #35 | 0.0403 | $0.0629 | $1.561 | |
| 53 raw #39 | 0.124 | $0.254139 | $2.056 | |
| 54 raw #41 | 0.0083 | $0.0189 | $2.277 | |
| 55 raw #42 | 0.118 | $0.28 | $2.371 | |
| 56 raw #43 | 0.0583 | $0.1424 | $2.443 | |
| 57 raw #46 | 0.0972 | $0.2642 | $2.718 | |
| 58 raw #47 | 0.0569 | $0.16 | $2.812 | |
| 59 raw #48 | 0.0778 | $0.22 | $2.828 | |
| 60 raw #49 | 0.267 | $0.7594 | $2.847 | |
| 61 raw #50 | 0.0403 | $0.12 | $2.978 | |
| 62 raw #51 | 0.0167 | $0.05 | $2.994 | |
| 63 raw #52 | 0.0194 | $0.06 | $3.093 | |
| 64 raw #53 | 0.0042 | $0.0131 | $3.119 | |
| 65 raw #54 | 0.433 | $1.393 | $3.215 | |
| 66 raw #56 | 0.0169 | $0.057 | $3.373 | |
| 67 raw #57 | 0.0694 | $0.2349 | $3.385 | |
| 68 raw #58 | 0.0125 | $0.0426 | $3.408 | |
| 69 raw #59 | 0.0375 | $0.1295 | $3.453 | |
| 70 raw #60 | 0.139 | $0.48 | $3.456 | |
| 71 raw #61 | 0.228 | $0.79 | $3.468 | |
| 72 raw #63 | 0.0361 | $0.13 | $3.601 | |
| 73 raw #64 | 0.0125 | $0.0471 | $3.768 | |
| 74 raw #67 | 0.189 | $0.75 | $3.968 | |
| 75 raw #68 | 0.132 | $0.56 | $4.246 | |
| 76 raw #70 | 0.0444 | $0.1977 | $4.453 | |
| 77 raw #73 | 0.0112 | $0.0527 | $4.705 | |
| 78 raw #74 | 0.0278 | $0.1392 | $5.007 | |
| 79 raw #75 | 0.0694 | $0.3499 | $5.042 | |
| 80 raw #76 | 0.0111 | $0.06 | $5.405 | |
| 81 raw #77 | 0.0167 | $0.0909 | $5.443 | |
| 82 raw #78 | 0.136 | $0.7589 | $5.576 | |
| 83 raw #79 | 0.0749 | $0.4486 | $5.989 | |
| 84 raw #80 | 0.013 | $0.08 | $6.154 | |
| 85 raw #81 | 0.376 | $2.4 | $6.376 | |
| 86 raw #82 | 0.0653 | $0.4206 | $6.441 | |
| 87 raw #83 | 0.176 | $1.1704 | $6.635 | |
| 88 raw #84 | 0.0194 | $0.129 | $6.649 | |
| 89 raw #85 | 0.0986 | $0.7302 | $7.406 | |
| 90 raw #86 | 0.0444 | $0.33 | $7.432 | |
| 91 raw #87 | 0.0593 | $0.4857 | $8.191 | |
| 92 raw #88 | 0.0403 | $0.3766 | $9.345 | |
| 93 raw #89 | 0.0212 | $0.1994 | $9.406 | |
| 94 raw #90 | 0.0194 | $0.1896 | $9.773 | |
| 95 raw #91 | 0.0236 | $0.2311 | $9.792 | |
| 96 raw #92 | 0.0127 | $0.1272 | $10.02 | |
| 97 raw #93 | 0.0199 | $0.2343 | $11.77 | |
| 98 raw #95 | 0.0212 | $0.2654 | $12.52 | |
| 99 raw #96 | 0.0254 | $0.3191 | $12.56 | |
| 100 raw #97 | 0.0653 | $0.8339 | $12.77 | |
| 101 raw #99 | 0.0208 | $0.2843 | $13.67 | |
| 102 raw #100 | 0.0042 | $0.0581 | $13.83 | |
| 103 raw #101 | 0.0611 | $0.856 | $14.01 | |
| 104 raw #102 | 0.0292 | $0.4439 | $15.2 | |
| 105 raw #103 | 0.0486 | $0.757 | $15.58 | |
| 106 raw #104 | 0.0198 | $0.3173 | $16.03 | |
| 107 raw #105 | 0.0298 | $0.4787 | $16.06 | |
| 108 raw #106 | 0.0042 | $0.0691 | $16.45 | |
| 109 raw #107 | 0.0085 | $0.1425 | $16.76 | |
| 110 raw #108 | 0.0403 | $0.7145 | $17.73 | |
| 111 raw #109 | 0.0299 | $0.5472 | $18.3 | |
| 112 raw #111 | 0.0861 | $1.9284 | $22.4 | |
| 113 raw #112 | 0.0083 | $0.1907 | $22.98 | |
| 114 raw #113 | 0.385 | $8.9928 | $23.38 | |
| 115 raw #114 | 0.0452 | $1.1569 | $25.6 | |
| 116 raw #116 | 0.004 | $0.14 | $35 | |
| 117 raw #117 | 0.183 | $7.1432 | $38.97 | |
| 118 raw #118 | 0.009 | $0.36 | $40 | |
| 119 raw #119 | 0.0127 | $0.6388 | $50.3 | |
| 120 raw #120 | 0.007 | $0.51 | $72.86 | |
| 121 raw #121 | 0.0205 | $2.2293 | $108.7 | |
| 122 raw #122 | 0.0486 | $7.5516 | $155.4 | |
| 123 raw #123 | 0.0194 | $4.7441 | $244.5 | |
| 124 raw #124 | 0.008 | $2.1 | $262.5 |
task cost divided by accuracy — expected spend per correct answer at this accuracy. 124 ranked, 1 retired excluded, 89 dropped for missing cost, 0 for a published cost of $0, 15 for a non-positive score
Ratio definition & methodology: 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. Fixed-capability half-life bands and threshold staircases are published on /methodology.
FRONTIER WATCH · top 24 of Epoch AI's Capability Index
every model near the top of Epoch AI's capability index, and what this archive holds for it on each tab. a tab ranks only configurations with a published score AND task cost; an empty cell means the measurer has not published that model on that benchmark yet. 24 of 24 are ranked on at least one tab.
| ECI # | Model | Lab | ECI | Code | Reason | Agent | IDE | ML-Code | ARC-2 |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Anthropic | 166.82 | #45 high | #17 max | — | #16 low | — | no cost |
| 2 | GPT-6 Astra | OpenAI | 166.41 | #91 max | #30 unknown | #7 low | — | #123 high | no cost |
| 3 | GPT-6.1 Sol | OpenAI | 166.32 | #61 max | #8 unknown | — | — | — | no cost |
| 4 | Claude Fable 5.1 | Anthropic | 164.65 | #93 high | #44 max | — | #47 low | #124 high | no cost |
| 5 | Gemini 4 Argon | Google DeepMind | 164.56 | — | #15 unknown | — | #33 minimal | — | — |
| 6 | Claude Sonnet 5.5 | Anthropic | 164.45 | #32 high | #12 max | — | #7 low | — | — |
| 7 | Claude Opus 5 | Anthropic | 162.6 | #90 high | #37 max | #11 low | #46 low | #106 | no cost |
| 8 | GPT-6 Sol | OpenAI | 162.55 | #75 max | #13 unknown | — | — | — | no cost |
| 9 | Claude Fable 5 | Anthropic | 162.13 | #92 high | #55 max | #31 low | — | #125 high | no cost |
| 10 | GPT-5.5 Pro | OpenAI | 161.97 | — | — | — | — | — | #94 high |
| 11 | GPT-5.6 Sol | OpenAI | 161.73 | #87 max | #31 max | #6 low | #19 low | #105 high | #5 medium |
| 12 | GPT-5.6 Terra | OpenAI | 159.43 | #84 max | #29 xhigh | #1 medium | #11 low | #66 high | #6 medium |
| 13 | GPT-5.4 Pro | OpenAI | 159.0 | — | — | — | — | #73 none | #110 xhigh |
| 14 | Claude Haiku 5.5 | Anthropic | 158.98 | — | #4 max | — | #2 low | — | — |
| 15 | GPT-5.5 | OpenAI | 158.95 | #34 none | #48 xhigh | #20 low | — | #92 none | #14 low |
| 16 | Claude Opus 4.8 | Anthropic | 158.26 | #70 none | #46 max | #29 low | — | #84 none | #45 low |
| 17 | Kimi K3 | Moonshot | 157.42 | #89 max | #36 unknown | #36 max | — | #115 max | #36 high |
| 18 | Gemini 3.7 Flash | Google DeepMind | 157.22 | #66 high | #18 unknown | #13 medium | — | — | no cost |
| 19 | Gemini 3.8 Flash | Google DeepMind | 157.2 | #67 high | #22 unknown | #10 medium | #42 medium | #76 high | no cost |
| 20 | Qwen 3.8 Max | Alibaba | 156.88 | — | #43 unknown | #32 xhigh | — | — | — |
| 21 | GPT-5.3 Codex | OpenAI | 156.76 | #73 xhigh | — | — | — | #120 | — |
| 22 | Muse Spark 1.3 | Meta AI | 156.76 | #46 xhigh | #14 unknown | — | #12 minimal | — | — |
| 23 | GPT-5.4 | OpenAI | 156.72 | #23 none | #54 xhigh | #51 xhigh | — | #72 none | #11 low |
| 24 | Claude Sonnet 5 | Anthropic | 156.36 | #86 high | #32 max | #39 low | #29 low | #104 high | — |
rule roster = top 24 models by ECI in the newest archived Epoch snapshot; identity = the versions Epoch scored each name under, resolved through the curated alias table (exact display-name match only where Epoch publishes no such link). source Epoch AI, epoch_capabilities_index/eci_scores.csv + processed_data_for_eci.csv (CC-BY-4.0)