OpenAI · model
GPT-5.2 benchmark scores
As of 2026-11-21, GPT-5.2 (OpenAI) has measured scores on 57 benchmarks and estimated scores on 122 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 90.3% | measured |
| AA-HLE | 37.7% | measured |
| AA-SciCode | 48.0% | estimated ± 3.6 pp, high confidence |
| AIME26 | 99.2% | estimated ± 1.2 pp, low confidence |
| Artificial Analysis Intelligence Index | 30.5% | measured |
| CritPt | 11.6% | measured |
| GDPval-AA | 36.8% | estimated ± 6.2 pp, medium confidence |
| GPQA Diamond | 92.4% | estimated ± 0.1 pp, high confidence |
| HLE | 44.8% | estimated ± 6.1 pp, medium confidence |
| IFBench | 73.6% | estimated ± 6.6 pp, medium confidence |
| AA-LCR | 82.7% | measured |
| LiveCodeBench v6 | 88.9% | estimated ± 2.2 pp, medium confidence |
| AA-Omniscience Accuracy | 44.3% | measured |
| SciCode | 47.5% | estimated ± 2.9 pp, medium confidence |
| BrowseComp | 65.8% | measured |
| HLE w/ tools | 38.4% | estimated ± 5.5 pp, medium confidence |
| IFEval | 93.0% | estimated ± 2.8 pp, high confidence |
| LongBench v2 | 68.1% | estimated ± 5.5 pp, medium confidence |
| VITA-Bench | 37.2% | estimated ± 6.7 pp, low confidence |
| τ²-bench results | 84.8% | measured |
| AA Coding Index | 60.9% | estimated ± 5.3 pp, medium confidence |
| AA-MMMU-Pro | 78.5% | estimated ± 1.0 pp, medium confidence |
| APEX-Agents-AA | 31.9% | estimated ± 3.6 pp, medium confidence |
| BioMysteryBench (human-difficult) | 44.9% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 78.4% | estimated ± 6.8 pp, low confidence |
| SWE-bench Verified | 80.0% | measured |
| BFCL v4 | 63.9% | estimated ± 8.5 pp, low confidence |
| CyberGym | 52.7% | estimated ± 7.3 pp, medium confidence |
| JobBench | 34.3% | measured |
| SkillsBench | 37.7% | estimated ± 4.1 pp, low confidence |
| SWE-bench Pro | 55.6% | measured |
| WideResearch | 77.8% | estimated ± 1.2 pp, medium confidence |
| AA Agentic Index | 14.4% | estimated ± 6.8 pp, medium confidence |
| MMLU-Pro | 85.2% | estimated ± 4.4 pp, high confidence |
| AA-IFBench | 75.4% | measured |
| AIME 2024 | 96.9% | estimated ± 2.6 pp, medium confidence |
| FailureSensorIQ | 67.6% | estimated ± 1.7 pp, high confidence |
| GPQA | 92.4% | measured |
| MMLU | 90.5% | estimated ± 0.9 pp, medium confidence |
| Vals GPQA Diamond | 91.7% | measured |
| Vals LiveCodeBench | 85.4% | measured |
| Vals MMLU-Pro | 86.2% | measured |
| MATH-500 | 99.4% | estimated ± 2.0 pp, medium confidence |
| MMMLU | 89.8% | estimated ± 0.7 pp, medium confidence |
| MMMU | 86.7% | measured |
| Chess Puzzles | 48.0% | measured |
| AIME 2025 | 100.0% | measured |
| AA MMLU-Pro | 87.4% | measured |
| Gert Labs | 46.5% | measured |
| AA EnterpriseOps-Gym | 44.7% | estimated ± 1.2 pp, medium confidence |
| AA Harvey LAB v1.0 | 86.9% | estimated ± 9.8 pp, low confidence |
| AA Tau3 Banking | 26.2% | estimated ± 3.6 pp, low confidence |
| ApprenticeBench | 5.0% | estimated ± 6.0 pp, low confidence |
| ARC-AGI-1 | 86.2% | measured |
| ARC-AGI-2 | 52.9% | measured |
| BabaIsBench | 95.9% | estimated ± 1.2 pp, high confidence |
| CursorBench 3.1 | 51.7% | estimated ± 5.6 pp, low confidence |
| CursorBench 3.2 | 51.4% | estimated ± 3.6 pp, high confidence |
| EBR-bench | 23.0% | measured |
| FrontierCode 1.1 Main | 27.5% | estimated ± 4.2 pp, medium confidence |
| FrontierMath v2 (Tier 4) | 31.7% | measured |
| FrontierMath v2 (Tiers 1-3) | 67.4% | measured |
| FrontierSWE v2 | 8.0% | estimated ± 8.3 pp, medium confidence |
| Furniture Assembly | 38.3% | measured |
| MaxxBench | 52.7% | estimated ± 4.3 pp, low confidence |
| MazeBench | 0.0% | estimated ± 5.1 pp, medium confidence |
| MirrorCode | 32.2% | measured |
| Mystery Game Puzzles | 23.0% | measured |
| NYT Connections | 83.6% | measured |
| OSWorld-Verified | 47.3% | measured |
| ParseBench-1 | 63.6% | estimated ± 8.4 pp, medium confidence |
| Terminal-Bench 4.1 | 2.2% | estimated ± 9.2 pp, low confidence |
| Terminal-Bench Hard | 47.0% | measured |
| Vals SWE-bench | 75.8% | measured |
| Terminal-Bench 2.1 (Vals AI) | 58.6% | estimated ± 5.6 pp, medium confidence |
| VulcanBench v3 | 83.9% | estimated ± 5.6 pp, low confidence |
| CWE-Bench-AA | 43.6% | estimated ± 6.4 pp, medium confidence |
| AA Cyber Index | 32.8% | estimated ± 13.5 pp, low confidence |
| DeepsecBench-AA | 4.5% | estimated ± 9.0 pp, low confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 55.1% | estimated ± 4.3 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 57.6% | estimated ± 0.8 pp, low confidence |
| AA ITBench | 37.0% | estimated ± 2.8 pp, high confidence |
| Bug Hunt Bench | 29.3% | estimated ± 12.1 pp, low confidence |
| CursorBench 4.0 | 32.3% | estimated ± 3.3 pp, medium confidence |
| DeepSWE | 56.3% | estimated ± 7.2 pp, low confidence |
| DRACO | 65.1% | estimated ± 2.9 pp, low confidence |
| FrontierCode 2.0 | 20.0% | estimated ± 6.7 pp, low confidence |
| FrontierPhysics | 2.4% | estimated ± 4.4 pp, high confidence |
| HLE w/o tools | 34.4% | estimated ± 4.9 pp, high confidence |
| OSWorld 2.0 | 0.0% | estimated ± 5.9 pp, low confidence |
| PostTrainBench v1.1 | 21.4% | estimated ± 1.3 pp, medium confidence |
| SpireClimb | 5.8% | estimated ± 7.2 pp, medium confidence |
| StealthMark | 46.1% | estimated ± 6.5 pp, medium confidence |
| Toolathlon-Verified | 50.5% | estimated ± 7.4 pp, medium confidence |
| VulcanBench Frontier v4 | 77.1% | estimated ± 8.3 pp, low confidence |
| GDP.pdf | 16.4% | estimated ± 4.1 pp, medium confidence |
| ARC-AGI-3 | 1.3% | estimated ± 3.1 pp, low confidence |
| ARC-AGI-3 Provider Adapter | 2.7% | estimated ± 7.3 pp, low confidence |
| SkillsBench official leaderboard (BenchFlow) | 51.7% | measured |
| BinaryAudit | 17.5% | measured |
| CompileBench | 91.1% | estimated ± 10.0 pp, low confidence |
| OTelBench | 26.1% | measured |
| Vibe Code Bench | 53.5% | measured |
| FACTS Multimodal | 47.0% | estimated ± 4.0 pp, high confidence |
| ParseBench-2 | 71.3% | estimated ± 5.0 pp, medium confidence |
| CharXiv | 82.1% | measured |
| CharXiv w/o tools | 83.6% | estimated ± 1.2 pp, high confidence |
| ExploitBench | 32.4% | estimated ± 6.5 pp, medium confidence |
| AI2D_TEST | 92.2% | measured |
| C-Eval | 94.7% | estimated ± 1.0 pp, low confidence |
| Claw-Eval | 63.3% | estimated ± 4.5 pp, medium confidence |
| CountBench | 91.9% | measured |
| DynaMath | 86.8% | measured |
| ERQA | 59.8% | measured |
| HMMT Feb 2025 | 96.6% | estimated ± 2.9 pp, medium confidence |
| HMMT Feb 2026 | 95.4% | estimated ± 1.5 pp, medium confidence |
| HMMT Nov 2025 | 95.4% | estimated ± 1.5 pp, low confidence |
| MathVision | 83.0% | measured |
| MCP Atlas | 63.6% | estimated ± 8.6 pp, medium confidence |
| MLVU (M-Avg) | 85.6% | measured |
| MMAnswerBench | 91.0% | estimated ± 2.4 pp, low confidence |
| MMLU-Redux | 96.5% | estimated ± 0.8 pp, low confidence |
| MMMU-Pro | 79.5% | measured |
| NL2Repo | 42.9% | estimated ± 4.7 pp, high confidence |
| OmniDocBench 1.5 | 85.7% | measured |
| PostTrainBench v1 | 21.4% | measured |
| RealWorldQA | 83.3% | measured |
| ScreenSpot Pro | 68.8% | estimated ± 4.8 pp, medium confidence |
| SimpleVQA | 55.8% | measured |
| SuperGPQA | 78.6% | estimated ± 8.5 pp, low confidence |
| Terminal-Bench 2.0 | 64.8% | estimated ± 5.4 pp, medium confidence |
| Toolathlon | 36.4% | estimated ± 4.2 pp, high confidence |
| V* | 75.9% | measured |
| Video-MME (w/o subtitle) | 85.8% | measured |
| Video-MME (with subtitle) | 86.0% | measured |
| VideoMMMU | 85.9% | measured |
| We-Math | 79.0% | measured |
| AA LiveCodeBench | 89.3% | estimated ± 2.6 pp, high confidence |
| HealthBench Hard | 32.1% | estimated ± 5.1 pp, low confidence |
| LiveCodeBench Pro | 80.2% | estimated ± 6.0 pp, low confidence |
| MedXpertQA (MM) | 73.5% | estimated ± 2.4 pp, high confidence |
| React Native Evals | 79.6% | estimated ± 2.7 pp, medium confidence |
| ResearchClawBench | 11.9% | estimated ± 2.1 pp, low confidence |
| SWE-Rebench | 61.8% | estimated ± 3.6 pp, medium confidence |
| PostTrainBench | 18.8% | estimated ± 1.2 pp, medium confidence |
| DAYJOB | 1.4% | estimated ± 2.7 pp, high confidence |
| DAYJOB Healthcare | 0.0% | estimated ± 4.0 pp, high confidence |
| BioMysteryBench (human-solvable) | 84.7% | estimated ± 0.6 pp, low confidence |
| FrontierCode 1.1 Extended | 54.5% | estimated ± 3.0 pp, low confidence |
| HealthBench (raw) | 46.6% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 53.6% | estimated ± 2.5 pp, low confidence |
| HealthBench Professional | 55.8% | estimated ± 4.2 pp, medium confidence |
| HealthBench Professional (raw) | 57.3% | estimated ± 4.5 pp, medium confidence |
| HLE-Verified | 20.0% | estimated ± 2.0 pp, low confidence |
| LABBench2 | 81.0% | estimated ± 1.8 pp, low confidence |
| ScarfBench | 46.7% | estimated ± 12.3 pp, low confidence |
| BlenderWorld (text-only) | 70.2% | estimated ± 2.8 pp, medium confidence |
| Cookie-Bench | 77.4% | estimated ± 1.3 pp, medium confidence |
| STEMBench | 2.5% | estimated ± 0.2 pp, medium confidence |
| SimpleQA | 51.5% | estimated ± 4.3 pp, low confidence |
| LiveCodeBench | 79.3% | estimated ± 14.6 pp, low confidence |
| Agents' Last Exam | 16.3% | estimated ± 11.6 pp, low confidence |
| BabyVision w/ Python | 88.8% | estimated ± 3.8 pp, low confidence |
| OpenHarmony Bench v1.0 | 53.0% | estimated ± 3.2 pp, high confidence |
| SEC-Bench Pro | 42.4% | estimated ± 7.5 pp, low confidence |
| BabyVision | 40.1% | estimated ± 3.7 pp, high confidence |
| MMVU | 78.1% | estimated ± 2.7 pp, medium confidence |
| PerceptionBench | 49.7% | estimated ± 3.6 pp, low confidence |
| ZeroBench | 15.7% | estimated ± 3.6 pp, low confidence |
| MRCRv2 | 85.9% | estimated ± 11.3 pp, low confidence |
| CC-OCR | 70.3% | measured |
| MMLongBench-Doc | 61.5% | estimated ± 2.6 pp, low confidence |
| RefCOCO (avg) | 90.1% | estimated ± 3.1 pp, medium confidence |
| OCRBench V2 | 71.5% | estimated ± 2.1 pp, medium confidence |
| BioBasicsBench | 31.2% | measured |
| LifeVal | 49.3% | measured |
| MMMU-Pro w/ Python | 81.5% | estimated ± 2.6 pp, high confidence |
| ExploitBench Internal Port | 1.4% | estimated ± 1.4 pp, medium confidence |
| PinchBench | 89.3% | estimated ± 3.6 pp, medium confidence |