OpenAI · model
GPT-6 Astra benchmark scores
As of 2026-10-07, GPT-6 Astra (OpenAI) has measured scores on 48 benchmarks and estimated scores on 75 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 96.1% | measured |
| AA-HLE | 54.7% | measured |
| AA-SciCode | 56.5% | measured |
| AIME26 | 100.0% | estimated ± 0.6 pp, low confidence |
| Artificial Analysis Intelligence Index | 52.7% | measured |
| CritPt | 31.7% | measured |
| GDPval-AA | 52.1% | measured |
| GPQA Diamond | 96.0% | measured |
| HLE | 60.9% | estimated ± 5.5 pp, low confidence |
| AA-LCR | 80.7% | measured |
| LiveCodeBench v6 | 93.6% | estimated ± 3.1 pp, medium confidence |
| AA-Omniscience Accuracy | 62.6% | measured |
| SciCode | 56.0% | estimated ± 3.2 pp, high confidence |
| Terminal-Bench 2.1 | 89.7% | estimated ± 3.9 pp, medium confidence |
| BrowseComp | 91.5% | measured |
| HLE w/ tools | 57.2% | measured |
| LongBench v2 | 61.5% | estimated ± 5.9 pp, low confidence |
| VITA-Bench | 46.5% | estimated ± 9.9 pp, low confidence |
| τ²-bench results | 85.6% | estimated ± 4.8 pp, medium confidence |
| AA Coding Index | 76.9% | measured |
| AA-MMMU-Pro | 86.9% | measured |
| APEX-Agents | 42.5% | estimated ± 10.0 pp, low confidence |
| APEX-Agents-AA | 38.9% | estimated ± 2.9 pp, low confidence |
| BioMysteryBench (human-difficult) | 50.2% | estimated ± 5.9 pp, low confidence |
| DeepSearchQA | 93.6% | estimated ± 4.5 pp, high confidence |
| SWE-bench Verified | 95.3% | estimated ± 0.8 pp, medium confidence |
| AutomationBench | 41.4% | measured |
| BFCL v4 | 77.6% | estimated ± 2.6 pp, medium confidence |
| CyberGym | 79.9% | estimated ± 2.4 pp, medium confidence |
| JobBench | 66.0% | estimated ± 4.9 pp, low confidence |
| SWE-bench Pro | 86.7% | estimated ± 5.1 pp, low confidence |
| WideResearch | 82.7% | estimated ± 3.4 pp, low confidence |
| AA Agentic Index | 51.5% | measured |
| MMLU-Pro | 87.6% | estimated ± 3.0 pp, medium confidence |
| FrontierMath v2 (Tier 4) | 97.6% | measured |
| FrontierMath v2 (Tiers 1-3) | 93.1% | estimated ± 7.4 pp, low confidence |
| GPQA | 96.0% | measured |
| Gert Labs | 100.0% | estimated ± 3.0 pp, low confidence |
| AA-AnalystAgent | 51.2% | measured |
| AA EnterpriseOps-Gym | 53.0% | estimated ± 1.8 pp, low confidence |
| AA Harvey LAB | 93.5% | estimated ± 1.2 pp, medium confidence |
| ApprenticeBench | 68.0% | measured |
| ARC-AGI-1 | 98.5% | measured |
| ARC-AGI-2 | 95.0% | measured |
| CursorBench 3.1 | 60.0% | estimated ± 3.4 pp, medium confidence |
| CursorBench 3.2 | 69.5% | estimated ± 1.6 pp, medium confidence |
| Terminal-Bench 3.0 | 38.5% | estimated ± 6.9 pp, low confidence |
| FrontierCode 1.1 Main | 53.3% | measured |
| FrontierSWE v2 | 65.5% | measured |
| OfficeQA Pro | 67.7% | estimated ± 7.2 pp, medium confidence |
| OSWorld-Verified | 99.7% | estimated ± 1.9 pp, low confidence |
| Vals GPQA Diamond | 96.1% | estimated ± 1.1 pp, high confidence |
| Vals LiveCodeBench | 89.0% | estimated ± 1.9 pp, high confidence |
| Vals MMLU-Pro | 88.9% | estimated ± 1.2 pp, medium confidence |
| Vals SWE-bench | 97.1% | estimated ± 1.4 pp, low confidence |
| Terminal-Bench 2.1 (Vals AI) | 87.3% | measured |
| VulcanBench v3 | 87.1% | estimated ± 0.7 pp, low confidence |
| AA AutomationBench | 68.5% | measured |
| GDP.pdf | 31.0% | measured |
| AA ITBench | 48.6% | measured |
| AA Tau3 Banking | 41.4% | measured |
| AA Terminal-Bench 2.1 | 88.4% | measured |
| AA Terminal-Bench 4.0 | 59.1% | measured |
| CursorBench 4.0 | 42.5% | estimated ± 1.6 pp, high confidence |
| CWE-bench v1 | 68.0% | measured |
| DeepSWE | 74.1% | measured |
| HLE w/o tools | 48.5% | estimated ± 4.4 pp, medium confidence |
| OSWorld 2.0 | 72.6% | measured |
| PostTrainBench v1.1 | 44.3% | measured |
| ProgramBench | 90.9% | estimated ± 4.8 pp, medium confidence |
| Terminal-Bench 4.0 | 57.9% | measured |
| Terminal-Bench-Science 0.1 | 64.6% | measured |
| Toolathlon-Verified | 80.0% | estimated ± 1.0 pp, low confidence |
| Vibe Code Bench | 80.8% | estimated ± 5.6 pp, low confidence |
| CharXiv | 88.5% | estimated ± 3.9 pp, medium confidence |
| CharXiv w/o tools | 86.7% | estimated ± 2.4 pp, medium confidence |
| ExploitBench | 100.0% | measured |
| ExploitGym | 42.4% | measured |
| C-Eval | 96.4% | estimated ± 0.9 pp, low confidence |
| Claw-Eval | 81.7% | estimated ± 7.9 pp, low confidence |
| HMMT Nov 2025 | 93.2% | estimated ± 2.6 pp, low confidence |
| MathVision | 96.2% | estimated ± 2.4 pp, low confidence |
| MCP Atlas | 87.3% | estimated ± 2.7 pp, low confidence |
| MMAnswerBench | 100.0% | estimated ± 1.3 pp, low confidence |
| MMLU-Redux | 97.9% | estimated ± 1.1 pp, low confidence |
| MMMU-Pro | 81.9% | estimated ± 2.7 pp, medium confidence |
| NL2Repo | 57.1% | estimated ± 5.1 pp, low confidence |
| QwenClawBench | 100.0% | estimated ± 6.0 pp, low confidence |
| ScreenSpot Pro | 92.7% | measured |
| SuperGPQA | 89.7% | estimated ± 7.5 pp, low confidence |
| Terminal-Bench 2.0 | 84.0% | estimated ± 5.9 pp, low confidence |
| Toolathlon | 66.4% | estimated ± 1.7 pp, low confidence |
| VideoMMMU | 90.6% | estimated ± 1.7 pp, low confidence |
| ERQA | 68.9% | estimated ± 3.9 pp, medium confidence |
| HealthBench Hard | 36.6% | measured |
| MedXpertQA (MM) | 77.8% | estimated ± 6.5 pp, low confidence |
| MedXpertQA (Text) | 77.7% | estimated ± 4.5 pp, low confidence |
| MMLU-Pro (Arcee) | 88.8% | estimated ± 3.4 pp, low confidence |
| React Native Evals | 83.4% | estimated ± 3.5 pp, medium confidence |
| ResearchClawBench | 21.1% | estimated ± 2.0 pp, medium confidence |
| SWE-Rebench | 58.1% | estimated ± 8.9 pp, low confidence |
| ARC-AGI-3 | 62.7% | measured |
| FrontierMath (legacy) | 90.7% | estimated ± 2.3 pp, low confidence |
| BioMysteryBench (human-solvable) | 89.5% | estimated ± 0.6 pp, medium confidence |
| DRACO | 82.9% | estimated ± 3.0 pp, medium confidence |
| FrontierCode 1.1 Extended | 64.5% | measured |
| HealthBench (raw) | 56.9% | measured |
| HealthBench (length-adjusted) | 58.3% | measured |
| HealthBench Professional | 64.7% | measured |
| HealthBench Professional (raw) | 68.2% | measured |
| HLE-Verified | 54.6% | estimated ± 0.7 pp, low confidence |
| LABBench2 | 85.6% | estimated ± 1.3 pp, medium confidence |
| MMMU | 93.3% | estimated ± 0.9 pp, medium confidence |
| LiveCodeBench | 100.0% | estimated ± 8.9 pp, low confidence |
| Agents' Last Exam | 59.3% | measured |
| OpenHarmony Bench | 58.8% | estimated ± 2.4 pp, high confidence |
| SEC-Bench Pro | 85.4% | measured |
| SimpleVQA | 58.1% | estimated ± 5.7 pp, low confidence |
| MMMLU | 91.6% | estimated ± 1.1 pp, low confidence |
| MMLU | 96.3% | estimated ± 1.1 pp, low confidence |
| MMMU-Pro w/ Python | 85.8% | estimated ± 1.6 pp, medium confidence |
| RealWorldQA | 91.0% | estimated ± 1.2 pp, low confidence |
| OmniDocBench 1.5 | 92.0% | estimated ± 6.4 pp, low confidence |