OpenAIGPT-6 Astra
Best configuration ranks #1 of 610 on the BenchLeader Index at 70.6 ±3.6 (max reasoning effort). Last measured 8 Sept 2026. Released 3 Sept 2026.
- Blended price
- $22.00/M
- $11.00 in · $55.00 out
- Output speed
- 54 tok/s
- First answer
- 329 s
- first token 7.00 s
- Context
- 1.1M
- Overall index71
- Reasoning71
- Coding79
- Agents & tools68
- Maths77
- Knowledge80
- Composite73
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 64.4 | #33 | 54 tok/s | 329 s | $0.021 | Coding 62 · Composite 86 · Knowledge 77 · Long context 62 · Maths 72 · Multimodal 67 · Reasoning 60 |
| low | 65.8 | #24 | 47 tok/s | 2.69 s | $0.019 | Agents & tools 56 · Coding 63 · Composite 87 · Knowledge 83 · Long context 67 · Maths 74 · Multimodal 69 · Reasoning 65 |
| medium | 67.4 | #15 | 44 tok/s | 5.42 s | $0.019 | Agents & tools 58 · Coding 63 · Composite 92 · Knowledge 84 · Long context 66 · Maths 79 · Multimodal 70 · Reasoning 69 |
| high | 69.5 | #9 | 49 tok/s | 93 s | $0.019 | Agents & tools 61 · Coding 72 · Composite 93 · Knowledge 85 · Long context 67 · Maths 79 · Multimodal 71 · Reasoning 71 |
| xhigh | 68.6 | #11 | 51 tok/s | 209 s | $0.019 | Agents & tools 61 · Coding 65 · Composite 95 · Knowledge 85 · Long context 67 · Maths 79 · Multimodal 71 · Reasoning 71 |
| maxbest | 70.6 | #1 | 54 tok/s | 329 s | $0.021 | Agents & tools 68 · Coding 79 · Composite 73 · Knowledge 80 · Maths 77 · Reasoning 71 |
| default | 69.9 | #7 | 54 tok/s | 329 s | $0.019 | Agents & tools 75 · Composite 95 · Knowledge 72 · Long context 67 · Maths 85 · Multimodal 72 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | – | – | – | 95.8%#1 | – | Epoch AI Benchmarking Hub | |
| LiveBench Reasoningnot in index | – | – | – | – | – | 92.7%#1 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 89.5%#74 | 93.1%#24 | 93.9%#10 | 95.0%#4 | 96.3%#1 | – | 96.1%#2 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 37.1%#71 | 49.2%#15 | 52.7%#12 | 53.1%#10 | 54.6%#7 | – | 54.7%#6 | Artificial Analysis | |
| ARC-AGI-1 | 86.0%#70 | 96.5%#13 | 97.5%#6 | 98.5%#1 | 98.5%#1 | 97.5%#6 | – | ARC Prize | |
| ARC-AGI-2 | 59.6%#60 | 85.4%#16 | 92.1%#4 | 92.1%#4 | 93.3%#2 | 95.0%#1 | – | ARC Prize | |
| ARC-AGI-3 | 35.2%#11 | 17.5%#13 | 38.6%#10 | 54.8%#9 | 59.3%#8 | 62.7%#7 | – | ARC Prize |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SciCode | 53.5%#41 | 54.0%#34 | 54.2%#33 | 55.4%#24 | 55.7%#21 | 56.5%#13 | – | SciCode | |
| WeirdML | – | – | – | 92.9%#2 | – | – | – | WeirdML | |
| FrontierCode | – | – | – | – | – | 53.3%#3 | – | Cognition | |
| LMArena WebDev | – | – | – | – | – | 1796#1 | – | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | – | 80.4%#13 | – | LiveBench | |
| SciCode (AA)not in index | 53.5%#47 | 54.0%#42 | 54.2%#41 | 55.4%#31 | 55.7%#28 | – | 56.5%#20 | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | 50.6%#25 | 54.2%#21 | 57.9%#16 | 57.9%#16 | 58.2%#15 | – | Terminal-Bench | |
| APEX-Agents | – | – | – | – | – | – | 46.7%#2 | Mercor | |
| LMArena Agent | – | – | – | – | – | 12.5#2 | – | LMArena | |
| LiveBench Agentic Codingnot in index | – | – | – | – | – | 57.3%#17 | – | LiveBench | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | – | 87.3%#1 | – | Vals AI |
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | – | 93.7%#1 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | 82.9%#8 | 87.8%#6 | 97.6%#4 | 97.6%#1 | 97.6%#1 | 97.6%#1 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | – | – | – | – | 100.0%#1 | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | – | – | 99.0%#2 | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | – | 96.8%#2 | – | LiveBench |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | – | 75.6%#1 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | – | 83.0%#1 | – | LiveBench | |
| AA-Omniscience | 26.6#27 | 40.5#9 | 42.2#7 | 43.7#1 | 43.4#3 | – | 43.4#4 | Artificial Analysis | |
| PRBench Finance | – | – | – | – | – | – | 47.5%#13 | Scale AI SEAL | |
| PRBench Legal | – | – | – | – | – | – | 48.4%#13 | Scale AI SEAL |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | – | 89.4%#3 | – | LiveBench |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| MMMU-Pro | 82.0%#22 | 84.6%#10 | 85.1%#6 | 86.4%#2 | 86.2%#3 | – | 86.9%#1 | Artificial Analysis |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| AA-LCR | 70.7%#154 | 80.0%#51 | 79.7%#59 | 80.0%#51 | 80.0%#51 | – | 80.7%#41 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | – | 166.6#1 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | – | 82.2%#3 | – | LiveBench | |
| AA Intelligence Index | 45.2#17 | 46.0#16 | 49.7#9 | 51.0#6 | 52.5#4 | – | 52.8#3 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | – | 66.6#3 | – | Vals AI |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
- #1GPQA Diamond95.8% · of 272
- #1FrontierMath Tiers 1–393.7% · of 103
- #1FrontierMath Tier 497.6% · of 59
- #1OTIS Mock AIME100.0% · of 255
- #1SimpleQA Verified75.6% · of 75
- #1LMArena WebDev1796 · of 118
- #1Terminal-Bench 2.1 (Vals)87.3% · of 61
- #1ARC-AGI-295.0% · of 187
- #2LMArena Agent12.5 · of 40
- #3FrontierCode53.3% · of 29
- #3LiveBench82.2% · of 52
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| OpenAI Flex | 55 tok/s | 3.69 s | $5.00 | $25.00 | 1.1M | – |
| OpenAI | 31 tok/s | 5.03 s | $10.00 | $50.00 | 1.1M | – |
| OpenAI Fast | 27 tok/s | 7.00 s | $20.00 | $100.00 | 1.1M | – |
| Azure | 9 tok/s | 15 s | $10.00 | $50.00 | 1.1M | – |
| Azure (US) | 3 tok/s | 66 s | $11.00 | $55.00 | 1.1M | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $1.10 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.021 | $0.018 | 5.6 min |
| Summarise a 30-page report | 12,000 / 600 | $0.165 | $0.076 | 5.7 min |
| Code edit | 6,000 / 1,500 | $0.148 | $0.104 | 5.9 min |
| Agentic coding session | 60,000 / 4,000 | $0.880 | $0.434 | 6.7 min |
| Structured extraction | 2,000 / 200 | $0.033 | $0.018 | 5.5 min |
See also
Data as of 9 Sept 2026. Compare these configurations.