OPENAI · MODEL TELEMETRY CARD
GPT-5.6 Luna
Canonical UID: openai/gpt-5.6-luna · Released: 2026-07-09
| Domain | Benchmark | Effort / Mode | ⚡ Value/Cost # | 🏆 Performance # | Score | Task $ | Unit Cost $ | Unit |
|---|---|---|---|---|---|---|---|---|
| Code | ALE-Bench | MAX | #8 / 94 | #13 / 94 | 1667.4 | $0.359171 | $0.0002154 | rating point |
| Reason | ProofBench | MAX | #5 / 62 | #18 / 62 | 0.6 | $0.123288 | $0.2055 | correct answer |
| Agent | DeepSWE | HIGH | #1 / 65 | #51 / 65 | 0.442 | $0.777901 | $1.758 | solved task |
| Agent | DeepSWE | MEDIUM | #58 / 65 | #64 / 65 | 0.113 | $0.21631 | $1.917 | solved task |
| Agent | DeepSWE | XHIGH | #6 / 65 | #35 / 65 | 0.569 | $1.53562 | $2.701 | solved task |
| Agent | DeepSWE | MAX | #18 / 65 | #20 / 65 | 0.672 | $3.02812 | $4.507 | solved task |
| Agent | DeepSWE | LOW | #60 / 65 | #65 / 65 | 0.0155 | $0.0724062 | $4.675 | solved task |
Compare against all models on ← #frontier Live Ranking Board · Fixed-capability cost half-life analysis on /methodology.