OpenAI · model
GPT-OSS 120B benchmark scores
As of 2026-11-22, GPT-OSS 120B (OpenAI) has measured scores on 25 benchmarks and estimated scores on 105 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 78.2% | measured |
| AA-HLE | 19.6% | measured |
| AA-SciCode | 34.0% | measured |
| AIME26 | 91.6% | estimated ± 4.3 pp, medium confidence |
| Artificial Analysis Intelligence Index | 3.8% | measured |
| CritPt | 1.1% | measured |
| GDPval-AA | 5.4% | measured |
| GPQA Diamond | 80.2% | estimated ± 1.2 pp, high confidence |
| HLE | 14.9% | measured |
| IFBench | 67.0% | estimated ± 6.6 pp, medium confidence |
| AA-LCR | 52.0% | measured |
| LiveCodeBench v6 | 78.5% | estimated ± 4.7 pp, high confidence |
| AA-Omniscience Accuracy | 21.7% | measured |
| SciCode | 34.3% | estimated ± 3.1 pp, high confidence |
| BrowseComp | 29.5% | estimated ± 4.2 pp, medium confidence |
| HLE w/ tools | 26.8% | estimated ± 4.1 pp, low confidence |
| IFEval | 92.8% | estimated ± 2.8 pp, high confidence |
| LongBench v2 | 52.5% | estimated ± 5.5 pp, medium confidence |
| τ²-bench results | 65.8% | measured |
| AA Coding Index | 30.4% | measured |
| APEX-Agents | 3.5% | estimated ± 9.4 pp, low confidence |
| APEX-Agents-AA | 3.1% | measured |
| BioMysteryBench (human-difficult) | 18.0% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 62.0% | estimated ± 4.5 pp, medium confidence |
| SWE-bench Verified | 62.4% | measured |
| BFCL v4 | 63.5% | estimated ± 2.7 pp, high confidence |
| CyberGym | 50.3% | estimated ± 5.3 pp, low confidence |
| JobBench | 0.0% | estimated ± 8.2 pp, low confidence |
| SWE-bench Pro | 39.0% | estimated ± 4.5 pp, high confidence |
| WideResearch | 63.1% | estimated ± 3.3 pp, medium confidence |
| AA Agentic Index | 6.2% | measured |
| MMLU-Pro | 81.9% | estimated ± 4.0 pp, high confidence |
| AA-IFBench | 69.0% | measured |
| AIME 2024 | 95.8% | measured |
| BBH | 95.6% | estimated ± 2.4 pp, low confidence |
| FailureSensorIQ | 50.5% | measured |
| GPQA | 80.1% | measured |
| MMLU | 90.0% | measured |
| Vals GPQA Diamond | 77.3% | estimated ± 3.2 pp, high confidence |
| Vals LiveCodeBench | 71.5% | estimated ± 5.7 pp, low confidence |
| Vals MMLU-Pro | 83.1% | estimated ± 2.3 pp, high confidence |
| MATH-500 | 98.2% | estimated ± 1.2 pp, medium confidence |
| MMMLU | 86.5% | estimated ± 0.7 pp, medium confidence |
| Chess Puzzles | 25.0% | estimated ± 4.7 pp, medium confidence |
| AIME 2025 | 92.5% | measured |
| AA MMLU-Pro | 84.3% | estimated ± 0.8 pp, medium confidence |
| Gert Labs | 21.1% | measured |
| AA Harvey LAB v1.0 | 79.4% | estimated ± 1.2 pp, low confidence |
| AA Tau3 Banking | 19.6% | estimated ± 3.7 pp, medium confidence |
| ApprenticeBench | 3.1% | estimated ± 10.4 pp, low confidence |
| ARC-AGI-1 | 51.6% | estimated ± 6.7 pp, low confidence |
| ARC-AGI-2 | 0.0% | estimated ± 10.9 pp, low confidence |
| CursorBench 3.2 | 28.1% | estimated ± 5.3 pp, low confidence |
| EBR-bench | 0.0% | estimated ± 10.4 pp, low confidence |
| FrontierSWE v2 | 2.1% | estimated ± 14.2 pp, low confidence |
| Furniture Assembly | 32.9% | estimated ± 9.3 pp, low confidence |
| MaxxBench | 47.2% | estimated ± 8.8 pp, low confidence |
| MazeBench | 0.0% | estimated ± 2.7 pp, medium confidence |
| Mystery Game Puzzles | 0.0% | measured |
| NYT Connections | 58.3% | estimated ± 13.8 pp, low confidence |
| OSWorld-Verified | 58.3% | estimated ± 8.4 pp, low confidence |
| Vals SWE-bench | 54.3% | estimated ± 5.5 pp, medium confidence |
| CWE-Bench-AA | 32.3% | estimated ± 8.1 pp, low confidence |
| AA Cyber Index | 4.5% | estimated ± 9.2 pp, low confidence |
| DeepsecBench-AA | 6.2% | estimated ± 8.2 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 4.7% | estimated ± 2.1 pp, low confidence |
| AA ITBench | 34.8% | estimated ± 4.0 pp, medium confidence |
| Bug Hunt Bench | 11.2% | estimated ± 10.5 pp, low confidence |
| CWE-bench v1 | 33.7% | estimated ± 5.7 pp, low confidence |
| DeepSWE | 15.9% | estimated ± 7.5 pp, low confidence |
| DRACO | 36.3% | estimated ± 2.0 pp, low confidence |
| FrontierPhysics | 2.5% | estimated ± 5.0 pp, low confidence |
| HLE w/o tools | 18.9% | estimated ± 4.9 pp, medium confidence |
| OSWorld 2.1 | 1.6% | estimated ± 3.6 pp, medium confidence |
| SpireClimb | 0.0% | estimated ± 9.3 pp, low confidence |
| StealthMark | 28.4% | estimated ± 9.2 pp, low confidence |
| Toolathlon-Verified | 20.9% | estimated ± 2.3 pp, low confidence |
| VulcanBench Frontier v4 | 62.6% | estimated ± 6.8 pp, low confidence |
| AA AutomationBench | 38.4% | estimated ± 9.9 pp, low confidence |
| GDP.pdf | 10.0% | estimated ± 4.2 pp, medium confidence |
| ARC-AGI-3 | 1.3% | estimated ± 9.8 pp, medium confidence |
| ARC-AGI-3 Provider Adapter | 2.7% | estimated ± 7.3 pp, low confidence |
| CWE-bench | 26.5% | estimated ± 6.2 pp, low confidence |
| MKA | 32.8% | estimated ± 4.6 pp, medium confidence |
| SkillsBench official leaderboard (BenchFlow) | 41.4% | estimated ± 6.3 pp, low confidence |
| CompileBench | 42.2% | measured |
| Vibe Code Bench | 27.7% | estimated ± 6.5 pp, low confidence |
| FACTS Multimodal | 37.5% | estimated ± 5.8 pp, low confidence |
| Hermes Bench | 22.4% | estimated ± 2.9 pp, medium confidence |
| Hermes Index | 15.0% | estimated ± 5.7 pp, low confidence |
| ExploitBench | 0.0% | estimated ± 7.2 pp, low confidence |
| ExploitGym | 0.0% | estimated ± 8.3 pp, low confidence |
| C-Eval | 86.0% | estimated ± 0.9 pp, low confidence |
| HMMT Feb 2025 | 95.5% | estimated ± 2.9 pp, high confidence |
| HMMT Feb 2026 | 78.0% | estimated ± 11.0 pp, low confidence |
| MCP Atlas | 64.6% | estimated ± 7.8 pp, low confidence |
| MMLU-Redux | 93.7% | estimated ± 0.8 pp, medium confidence |
| NL2Repo | 19.9% | estimated ± 4.7 pp, medium confidence |
| QwenClawBench | 55.9% | estimated ± 6.0 pp, low confidence |
| SuperGPQA | 56.3% | estimated ± 7.2 pp, medium confidence |
| Toolathlon | 27.0% | estimated ± 6.3 pp, low confidence |
| AA LiveCodeBench | 87.8% | measured |
| HealthBench Hard | 10.4% | estimated ± 8.6 pp, low confidence |
| MMLU-Pro (Arcee) | 81.3% | estimated ± 4.0 pp, medium confidence |
| React Native Evals | 71.6% | measured |
| ResearchClawBench | 9.0% | estimated ± 1.9 pp, medium confidence |
| SWE-Rebench | 57.0% | estimated ± 3.6 pp, low confidence |
| DAYJOB | 1.3% | estimated ± 3.2 pp, medium confidence |
| DAYJOB Healthcare | 0.7% | estimated ± 4.2 pp, medium confidence |
| Finance Agent v2 | 55.1% | estimated ± 2.4 pp, low confidence |
| BioMysteryBench (human-solvable) | 81.7% | estimated ± 0.6 pp, low confidence |
| HealthBench (raw) | 27.3% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 45.8% | estimated ± 3.4 pp, low confidence |
| HealthBench Professional | 56.7% | estimated ± 4.2 pp, low confidence |
| HealthBench Professional (raw) | 39.3% | estimated ± 4.5 pp, medium confidence |
| HLE-Verified | 16.9% | estimated ± 7.1 pp, low confidence |
| ScarfBench | 24.4% | estimated ± 9.1 pp, low confidence |
| BlenderWorld (text-only) | 70.0% | estimated ± 2.8 pp, low confidence |
| Cookie-Bench | 76.9% | estimated ± 1.3 pp, low confidence |
| STEMBench | 2.3% | estimated ± 0.2 pp, low confidence |
| SimpleQA | 20.4% | estimated ± 4.3 pp, medium confidence |
| LiveCodeBench | 67.7% | estimated ± 14.6 pp, low confidence |
| Agents' Last Exam | 29.6% | estimated ± 9.7 pp, low confidence |
| OpenHarmony Bench v1.0 | 51.0% | estimated ± 2.4 pp, medium confidence |
| SEC-Bench Pro | 29.0% | estimated ± 9.3 pp, low confidence |
| MRCRv2 | 23.0% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 59.6% | estimated ± 2.6 pp, low confidence |
| BioBasicsBench | 34.1% | estimated ± 3.8 pp, low confidence |
| LifeVal | 0.9% | estimated ± 5.0 pp, low confidence |
| ExploitBench Internal Port | 0.0% | estimated ± 5.2 pp, low confidence |