OpenAI · model
o3 benchmark scores
As of 2026-11-21, o3 (OpenAI) has measured scores on 16 benchmarks and estimated scores on 97 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 82.7% | measured |
| AA-HLE | 20.1% | measured |
| AA-SciCode | 51.8% | estimated ± 5.1 pp, low confidence |
| AIME26 | 95.2% | estimated ± 1.4 pp, high confidence |
| Artificial Analysis Intelligence Index | 18.4% | estimated ± 3.6 pp, high confidence |
| CritPt | 1.1% | measured |
| GDPval-AA | 13.8% | estimated ± 11.0 pp, low confidence |
| GPQA Diamond | 87.5% | estimated ± 1.8 pp, high confidence |
| HLE | 23.4% | estimated ± 8.5 pp, medium confidence |
| IFBench | 69.4% | estimated ± 6.6 pp, medium confidence |
| AA-LCR | 74.7% | measured |
| LiveCodeBench v6 | 88.2% | estimated ± 2.2 pp, medium confidence |
| AA-Omniscience Accuracy | 29.4% | estimated ± 4.8 pp, high confidence |
| SciCode | 44.5% | estimated ± 2.9 pp, low confidence |
| BrowseComp | 67.1% | estimated ± 13.0 pp, low confidence |
| HLE w/ tools | 32.9% | estimated ± 5.5 pp, low confidence |
| IFEval | 92.9% | estimated ± 2.8 pp, high confidence |
| LongBench v2 | 61.5% | estimated ± 5.5 pp, medium confidence |
| τ²-bench results | 80.7% | measured |
| AA Coding Index | 63.1% | estimated ± 11.5 pp, low confidence |
| AA-MMMU-Pro | 70.1% | measured |
| APEX-Agents-AA | 1.6% | estimated ± 11.4 pp, low confidence |
| BioMysteryBench (human-difficult) | 18.0% | estimated ± 6.8 pp, low confidence |
| SWE-bench Verified | 81.4% | estimated ± 13.1 pp, low confidence |
| BFCL v4 | 61.8% | estimated ± 8.5 pp, low confidence |
| SWE-bench Pro | 57.2% | estimated ± 4.7 pp, high confidence |
| AA Agentic Index | 19.7% | estimated ± 12.7 pp, low confidence |
| MMLU-Pro | 85.2% | estimated ± 4.8 pp, high confidence |
| AA-IFBench | 71.4% | measured |
| FailureSensorIQ | 67.6% | measured |
| GPQA | 86.0% | estimated ± 4.5 pp, high confidence |
| MMLU | 93.1% | estimated ± 1.2 pp, medium confidence |
| Vals GPQA Diamond | 84.1% | measured |
| Vals LiveCodeBench | 83.9% | measured |
| Vals MMLU-Pro | 85.6% | measured |
| MMMLU | 86.5% | estimated ± 0.7 pp, medium confidence |
| MMMU | 80.2% | estimated ± 2.7 pp, high confidence |
| Chess Puzzles | 36.0% | measured |
| AA MMLU-Pro | 85.3% | measured |
| ARC-AGI-1 | 76.3% | estimated ± 6.5 pp, medium confidence |
| ARC-AGI-2 | 33.5% | estimated ± 13.8 pp, low confidence |
| CursorBench 3.2 | 59.1% | estimated ± 7.2 pp, low confidence |
| EBR-bench | 18.9% | estimated ± 10.4 pp, low confidence |
| FrontierCode 1.1 Main | 38.0% | estimated ± 6.2 pp, medium confidence |
| FrontierMath v2 (Tier 4) | 4.8% | estimated ± 4.0 pp, high confidence |
| FrontierMath v2 (Tiers 1-3) | 33.3% | measured |
| Furniture Assembly | 33.5% | estimated ± 9.3 pp, medium confidence |
| MaxxBench | 47.8% | estimated ± 8.8 pp, medium confidence |
| Mystery Game Puzzles | 29.0% | measured |
| NYT Connections | 79.5% | estimated ± 12.6 pp, low confidence |
| OfficeQA Pro | 50.3% | estimated ± 6.8 pp, low confidence |
| ParseBench-1 | 53.9% | estimated ± 6.2 pp, medium confidence |
| Vals SWE-bench | 52.9% | estimated ± 9.0 pp, medium confidence |
| VulcanBench v3 | 82.9% | estimated ± 5.6 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 5.5% | estimated ± 2.1 pp, low confidence |
| Chartography (tools) | 77.0% | estimated ± 4.1 pp, low confidence |
| CursorBench 4.0 | 40.7% | estimated ± 4.5 pp, high confidence |
| FrontierPhysics | 2.7% | estimated ± 5.0 pp, medium confidence |
| HLE w/o tools | 32.3% | estimated ± 4.9 pp, high confidence |
| SpireClimb | 0.1% | estimated ± 9.3 pp, medium confidence |
| StealthMark | 43.6% | estimated ± 9.2 pp, medium confidence |
| ARC-AGI-3 | 0.0% | estimated ± 8.6 pp, low confidence |
| ARC-AGI-3 Provider Adapter | 2.7% | estimated ± 7.3 pp, low confidence |
| SkillsBench official leaderboard (BenchFlow) | 55.3% | estimated ± 6.2 pp, medium confidence |
| BinaryAudit | 35.0% | estimated ± 8.9 pp, low confidence |
| CompileBench | 100.0% | estimated ± 10.0 pp, low confidence |
| OTelBench | 31.2% | estimated ± 4.1 pp, low confidence |
| Vibe Code Bench | 0.0% | estimated ± 6.6 pp, low confidence |
| FACTS Multimodal | 42.1% | estimated ± 5.6 pp, medium confidence |
| ParseBench-2 | 58.1% | estimated ± 6.2 pp, medium confidence |
| CharXiv | 77.7% | estimated ± 6.3 pp, low confidence |
| CharXiv w/o tools | 77.0% | estimated ± 2.5 pp, medium confidence |
| ExploitBench | 49.1% | estimated ± 7.2 pp, medium confidence |
| AI2D_TEST | 88.1% | estimated ± 2.0 pp, medium confidence |
| C-Eval | 91.9% | estimated ± 1.1 pp, medium confidence |
| CountBench | 91.6% | estimated ± 2.9 pp, low confidence |
| DynaMath | 71.6% | estimated ± 3.0 pp, low confidence |
| ERQA | 43.9% | estimated ± 3.1 pp, medium confidence |
| HMMT Feb 2026 | 86.2% | estimated ± 3.1 pp, high confidence |
| HMMT Nov 2025 | 92.6% | estimated ± 1.5 pp, medium confidence |
| MathVision | 78.1% | estimated ± 3.1 pp, high confidence |
| MMAnswerBench | 82.3% | estimated ± 2.4 pp, medium confidence |
| MMLU-Redux | 93.6% | estimated ± 3.3 pp, high confidence |
| MMMU-Pro | 71.5% | estimated ± 1.5 pp, high confidence |
| NL2Repo | 45.7% | estimated ± 7.9 pp, low confidence |
| RealWorldQA | 79.8% | estimated ± 2.9 pp, low confidence |
| ScreenSpot Pro | 71.3% | estimated ± 10.8 pp, low confidence |
| V* | 51.0% | estimated ± 7.5 pp, low confidence |
| Video-MME (w/o subtitle) | 79.8% | estimated ± 1.7 pp, medium confidence |
| Video-MME (with subtitle) | 81.8% | estimated ± 3.7 pp, low confidence |
| AA LiveCodeBench | 80.8% | measured |
| HealthBench Hard | 10.4% | estimated ± 8.6 pp, low confidence |
| MedXpertQA (MM) | 51.5% | estimated ± 6.5 pp, low confidence |
| MMLU-Pro (Arcee) | 81.3% | estimated ± 4.0 pp, medium confidence |
| DAYJOB | 1.3% | estimated ± 3.2 pp, high confidence |
| DAYJOB Healthcare | 0.7% | estimated ± 4.2 pp, high confidence |
| BioMysteryBench (human-solvable) | 81.7% | estimated ± 0.6 pp, low confidence |
| HealthBench (raw) | 27.8% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 46.0% | estimated ± 3.4 pp, low confidence |
| HealthBench Professional | 58.1% | estimated ± 4.5 pp, low confidence |
| HealthBench Professional (raw) | 39.9% | estimated ± 4.5 pp, medium confidence |
| HLE-Verified | 0.2% | estimated ± 4.6 pp, low confidence |
| LABBench2 | 80.7% | estimated ± 1.8 pp, low confidence |
| BlenderWorld (text-only) | 70.7% | estimated ± 2.8 pp, medium confidence |
| Cookie-Bench | 78.8% | estimated ± 1.3 pp, medium confidence |
| SimpleQA | 30.0% | estimated ± 7.8 pp, low confidence |
| BabyVision w/ Python | 84.3% | estimated ± 4.0 pp, low confidence |
| SEC-Bench Pro | 48.5% | estimated ± 9.3 pp, medium confidence |
| MRCRv2 | 83.9% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 61.2% | estimated ± 2.6 pp, low confidence |
| BioBasicsBench | 33.9% | estimated ± 10.7 pp, low confidence |
| MMMU-Pro w/ Python | 73.8% | estimated ± 2.0 pp, high confidence |
| ExploitBench Internal Port | 0.2% | estimated ± 1.4 pp, medium confidence |