Leaderboard / Model
GPT-6 Astra
OpenAI · Closed weights · Released Sep 3, 2026 · $10 in · $50 out per 1M tokens
87Overall · rank 3
Score by job
Coding
81 · rank 4 · 3 of 5 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| DeepSWE | 74.1% | gpt-6-astra_xhigh | 1.09 |
| FrontierCode | 53.3% | gpt-6-astra_max | 1.42 |
| LMArena Coding | 1489 rating | gpt-6-astra-max | -0.10 |
Agents
100 · rank 1 · 2 of 4 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| APEX-Agents | 64.7% | gpt-6-astra_unknown | 1.18 |
| Remote Labor Index | 20.8% | gpt-6-astra_unknown | 1.83 |
Reasoning
100 · rank 1 · 3 of 4 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| GPQA Diamond | 95.8% | gpt-6-astra_max | 1.70 |
| ARC-AGI-2 | 95.0% | gpt-6-astra_max | 1.43 |
| Humanity’s Last Exam | 54.8% | gpt-6-astra_unknown | 1.96 |
Math
100 · rank 1 · 4 of 5 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| FrontierMath Tiers 1–3 | 93.7% | gpt-6-astra_max | 1.77 |
| FrontierMath Tier 4 | 97.6% | gpt-6-astra_high | 2.37 |
| OTIS Mock AIME | 100.0% | gpt-6-astra_max | 1.09 |
| ProofBench | 99.0% | gpt-6-astra_unknown | 2.02 |
Writing
46 · rank 36 · 2 of 2 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| LMArena Creative Writing | 1429 rating | gpt-6-astra-max | -0.68 |
| LMArena Instruction Following | 1439 rating | gpt-6-astra-max | -0.94 |
Design
94 · rank 2 · 1 of 1 benchmarks
| Benchmark | Result | Variant | z |
|---|---|---|---|
| LMArena WebDev | 1792 rating | gpt-6-astra-max | 2.32 |