OPENAI · MODEL TELEMETRY CARD
GPT-6 Astra
Canonical UID: openai/gpt-6-astra · Released: 2026-09-03
| Domain | Benchmark | Effort / Mode | ⚡ Value/Cost # | 🏆 Performance # | Score | Task $ | Unit Cost $ | Unit |
|---|---|---|---|---|---|---|---|---|
| Code | ALE-Bench | MAX | #19 / 94 | #1 / 94 | 2951.32 | $2.86415 | $0.0009705 | rating point |
| Reason | ProofBench | STD | #18 / 62 | #6 / 62 | 0.99 | $1.71238 | $1.73 | correct answer |
| Agent | DeepSWE | LOW | #4 / 65 | #21 / 65 | 0.67 | $1.59522 | $2.38 | solved task |
| Agent | DeepSWE | MEDIUM | #15 / 65 | #8 / 65 | 0.728 | $3.07547 | $4.225 | solved task |
| Agent | DeepSWE | HIGH | #23 / 65 | #4 / 65 | 0.732 | $3.92372 | $5.358 | solved task |
| Agent | DeepSWE | XHIGH | #25 / 65 | #1 / 65 | 0.741 | $4.42912 | $5.976 | solved task |
| Agent | DeepSWE | MAX | #40 / 65 | #5 / 65 | 0.732 | $7.49783 | $10.24 | solved task |
Compare against all models on ← #frontier Live Ranking Board · Fixed-capability cost half-life analysis on /methodology.