Google · model
Gemini 3.1 Pro benchmark scores
As of 2026-11-21, Gemini 3.1 Pro (Google) has measured scores on 60 benchmarks and estimated scores on 135 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 94.1% | measured |
| AA-HLE | 47.0% | measured |
| AA-SciCode | 58.7% | measured |
| AIME26 | 97.7% | estimated ± 1.2 pp, medium confidence |
| Artificial Analysis Intelligence Index | 36.2% | estimated ± 5.6 pp, medium confidence |
| CritPt | 17.7% | measured |
| GDPval-AA | 14.6% | measured |
| GPQA Diamond | 94.3% | measured |
| HLE | 54.3% | estimated ± 2.2 pp, high confidence |
| IFBench | 75.5% | estimated ± 6.6 pp, medium confidence |
| AA-LCR | 82.0% | measured |
| LiveCodeBench v6 | 90.3% | estimated ± 2.2 pp, low confidence |
| AA-Omniscience Accuracy | 58.4% | estimated ± 6.5 pp, medium confidence |
| SciCode | 53.8% | estimated ± 2.9 pp, low confidence |
| Terminal-Bench 2.1 | 73.0% | estimated ± 5.6 pp, medium confidence |
| BrowseComp | 84.5% | estimated ± 4.2 pp, high confidence |
| HLE w/ tools | 56.9% | estimated ± 4.1 pp, medium confidence |
| IFEval | 93.1% | estimated ± 2.8 pp, high confidence |
| LongBench v2 | 67.4% | estimated ± 5.5 pp, medium confidence |
| VITA-Bench | 35.6% | estimated ± 10.7 pp, low confidence |
| τ²-bench results | 95.6% | measured |
| AA Coding Index | 68.8% | measured |
| AA-MMMU-Pro | 82.4% | measured |
| APEX-Agents | 33.2% | estimated ± 4.9 pp, low confidence |
| APEX-Agents-AA | 32.0% | measured |
| BioMysteryBench (human-difficult) | 47.5% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 69.7% | measured |
| SWE-bench Verified | 82.9% | estimated ± 3.2 pp, medium confidence |
| BFCL v4 | 70.1% | estimated ± 2.7 pp, high confidence |
| CyberGym | 53.5% | estimated ± 5.3 pp, low confidence |
| JobBench | 39.9% | estimated ± 8.2 pp, low confidence |
| SWE-bench Pro | 55.5% | estimated ± 2.9 pp, medium confidence |
| WideResearch | 77.3% | estimated ± 5.9 pp, low confidence |
| AA Agentic Index | 10.3% | measured |
| MMLU-Pro | 85.7% | estimated ± 2.5 pp, low confidence |
| AA-IFBench | 77.1% | measured |
| FailureSensorIQ | 68.6% | estimated ± 1.7 pp, high confidence |
| GPQA | 94.3% | estimated ± 0.0 pp, high confidence |
| MMLU | 96.9% | estimated ± 1.2 pp, medium confidence |
| Vals GPQA Diamond | 95.5% | measured |
| Vals LiveCodeBench | 88.5% | measured |
| Vals MMLU-Pro | 91.0% | measured |
| MMMLU | 90.6% | estimated ± 0.7 pp, low confidence |
| MMMU | 88.0% | estimated ± 0.8 pp, medium confidence |
| Chess Puzzles | 51.0% | measured |
| AA MMLU-Pro | 89.6% | estimated ± 1.1 pp, medium confidence |
| Gert Labs | 56.9% | measured |
| AA-AnalystAgent | 51.6% | estimated ± 5.9 pp, low confidence |
| AA EnterpriseOps-Gym | 42.2% | measured |
| AA Harvey LAB v1.0 | 79.4% | estimated ± 1.2 pp, low confidence |
| AA Tau3 Banking | 39.2% | estimated ± 2.7 pp, medium confidence |
| AA Terminal-Bench 2.1 | 82.9% | estimated ± 2.5 pp, high confidence |
| ApprenticeBench | 0.0% | estimated ± 3.1 pp, low confidence |
| ARC-AGI-1 | 94.1% | estimated ± 1.7 pp, high confidence |
| ARC-AGI-2 | 77.1% | measured |
| BabaIsBench | 92.0% | measured |
| CursorBench 3.1 | 65.5% | estimated ± 3.4 pp, medium confidence |
| CursorBench 3.2 | 55.8% | estimated ± 4.4 pp, high confidence |
| EBR-bench | 14.3% | measured |
| Terminal-Bench 3.0 | 8.9% | estimated ± 3.2 pp, medium confidence |
| FrontierCode 1.1 Main | 40.7% | estimated ± 2.8 pp, low confidence |
| FrontierMath v2 (Tier 4) | 26.8% | measured |
| FrontierMath v2 (Tiers 1-3) | 59.6% | measured |
| FrontierSWE v2 | 9.7% | estimated ± 9.5 pp, medium confidence |
| Furniture Assembly | 26.7% | measured |
| MaxxBench | 42.0% | estimated ± 4.3 pp, low confidence |
| MazeBench | 1.0% | measured |
| MirrorCode | 8.9% | measured |
| Mystery Game Puzzles | 34.0% | measured |
| NYT Connections | 97.4% | measured |
| OfficeQA Pro | 67.3% | estimated ± 6.8 pp, medium confidence |
| OSWorld-Verified | 67.9% | estimated ± 3.8 pp, medium confidence |
| ParseBench-1 | 67.3% | measured |
| Terminal-Bench 4.1 | 2.7% | measured |
| Terminal-Bench Hard | 53.8% | measured |
| Vals SWE-bench | 78.8% | measured |
| Terminal-Bench 2.1 (Vals AI) | 70.8% | measured |
| VulcanBench v3 | 86.2% | estimated ± 5.6 pp, low confidence |
| CWE-Bench-AA | 20.9% | estimated ± 10.1 pp, low confidence |
| AA Cyber Index | 7.3% | estimated ± 9.2 pp, low confidence |
| DeepsecBench-AA | 0.1% | estimated ± 8.9 pp, low confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 58.1% | estimated ± 4.4 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 63.9% | estimated ± 0.8 pp, low confidence |
| AA ITBench | 39.3% | estimated ± 2.8 pp, high confidence |
| Bug Hunt Bench | 5.0% | estimated ± 6.3 pp, low confidence |
| Chartography (tools) | 84.8% | estimated ± 4.1 pp, low confidence |
| CursorBench 4.0 | 34.1% | estimated ± 2.0 pp, medium confidence |
| CWE-bench v1 | 57.4% | estimated ± 5.1 pp, medium confidence |
| DeepSWE | 54.4% | estimated ± 4.3 pp, medium confidence |
| DRACO | 86.9% | estimated ± 0.7 pp, low confidence |
| FrontierCode 2.0 | 19.1% | estimated ± 6.0 pp, low confidence |
| FrontierPhysics | 0.0% | estimated ± 4.4 pp, high confidence |
| HLE w/o tools | 45.4% | measured |
| OSWorld 2.0 | 5.9% | estimated ± 5.9 pp, low confidence |
| OSWorld 2.1 | 78.6% | estimated ± 6.1 pp, medium confidence |
| OSWorld 2.1 Offline | 68.5% | estimated ± 13.8 pp, low confidence |
| PostTrainBench v1.1 | 22.0% | measured |
| ProgramBench | 0.0% | estimated ± 4.8 pp, low confidence |
| SpireClimb | 16.8% | estimated ± 7.2 pp, medium confidence |
| StealthMark | 40.1% | estimated ± 2.8 pp, medium confidence |
| SWE-bench Multimodal | 55.7% | estimated ± 5.6 pp, low confidence |
| Terminal-Bench 4.0 | 11.5% | estimated ± 4.5 pp, medium confidence |
| Toolathlon-Verified | 31.1% | estimated ± 2.3 pp, low confidence |
| VulcanBench Frontier v4 | 68.9% | estimated ± 2.8 pp, low confidence |
| AA AutomationBench | 55.0% | estimated ± 12.8 pp, low confidence |
| GDP.pdf | 17.0% | estimated ± 4.6 pp, medium confidence |
| AA Terminal-Bench 4.0 | 2.3% | estimated ± 6.2 pp, low confidence |
| ARC-AGI-3 | 0.4% | measured |
| ARC-AGI-3 Provider Adapter | 2.2% | estimated ± 4.4 pp, high confidence |
| cursorBench41 | 45.0% | estimated ± 3.7 pp, high confidence |
| CWE-bench | 54.7% | estimated ± 7.9 pp, low confidence |
| MKA | 42.4% | estimated ± 5.7 pp, low confidence |
| SkillsBench official leaderboard (BenchFlow) | 60.8% | measured |
| BinaryAudit | 49.1% | measured |
| CompileBench | 100.0% | estimated ± 10.0 pp, low confidence |
| OTelBench | 20.1% | estimated ± 5.6 pp, low confidence |
| Vibe Code Bench | 32.0% | measured |
| FACTS Multimodal | 50.4% | measured |
| Hermes Bench | 43.2% | estimated ± 2.9 pp, medium confidence |
| Hermes Index | 37.2% | estimated ± 3.1 pp, high confidence |
| ParseBench-2 | 73.3% | estimated ± 3.5 pp, high confidence |
| CharXiv | 80.2% | measured |
| CharXiv w/o tools | 88.9% | estimated ± 1.2 pp, high confidence |
| ExploitBench | 31.5% | estimated ± 1.9 pp, low confidence |
| ExploitGym | 6.6% | estimated ± 2.7 pp, medium confidence |
| AI2D_TEST | 93.4% | estimated ± 0.9 pp, medium confidence |
| C-Eval | 94.9% | estimated ± 1.2 pp, low confidence |
| Claw-Eval | 57.8% | measured |
| CountBench | 97.3% | estimated ± 2.2 pp, medium confidence |
| DynaMath | 85.7% | estimated ± 0.8 pp, medium confidence |
| ERQA | 69.4% | measured |
| HMMT Feb 2026 | 93.6% | estimated ± 1.5 pp, medium confidence |
| HMMT Nov 2025 | 95.0% | estimated ± 1.5 pp, low confidence |
| MathVision | 100.0% | estimated ± 2.2 pp, low confidence |
| MCP Atlas | 64.6% | estimated ± 7.8 pp, low confidence |
| MCP-Tasks | 69.3% | estimated ± 6.4 pp, low confidence |
| MLVU (M-Avg) | 85.7% | estimated ± 1.5 pp, high confidence |
| MMAnswerBench | 89.0% | estimated ± 2.4 pp, low confidence |
| MMLU-Redux | 93.2% | estimated ± 2.8 pp, medium confidence |
| MMMU-Pro | 83.9% | measured |
| NL2Repo | 44.7% | estimated ± 4.7 pp, high confidence |
| OmniDocBench 1.5 | 90.0% | estimated ± 2.0 pp, medium confidence |
| PostTrainBench v1 | 21.6% | measured |
| QwenClawBench | 55.9% | estimated ± 6.0 pp, low confidence |
| RealWorldQA | 87.7% | estimated ± 1.4 pp, medium confidence |
| ScreenSpot Pro | 84.4% | measured |
| SimpleVQA | 72.4% | measured |
| SuperGPQA | 83.0% | estimated ± 13.9 pp, low confidence |
| Terminal-Bench 2.0 | 70.7% | estimated ± 5.4 pp, medium confidence |
| Toolathlon | 27.0% | estimated ± 6.3 pp, low confidence |
| V* | 96.1% | estimated ± 6.5 pp, low confidence |
| Video-MME (w/o subtitle) | 87.4% | estimated ± 1.3 pp, medium confidence |
| Video-MME (with subtitle) | 86.6% | estimated ± 2.9 pp, high confidence |
| VideoMMMU | 85.3% | estimated ± 1.5 pp, medium confidence |
| We-Math | 87.2% | estimated ± 2.1 pp, medium confidence |
| AA LiveCodeBench | 90.9% | estimated ± 2.6 pp, medium confidence |
| HealthBench Hard | 20.6% | measured |
| LiveCodeBench Pro | 82.9% | measured |
| MedXpertQA (MM) | 81.3% | measured |
| MedXpertQA (Text) | 71.5% | measured |
| MMLU-Pro (Arcee) | 88.3% | estimated ± 3.4 pp, low confidence |
| React Native Evals | 78.9% | measured |
| ResearchClawBench | 13.3% | measured |
| SWE-Rebench | 58.9% | estimated ± 6.5 pp, low confidence |
| PostTrainBench | 18.1% | measured |
| DAYJOB | 0.0% | measured |
| DAYJOB Healthcare | 0.0% | measured |
| Finance Agent v2 | 56.5% | estimated ± 3.8 pp, medium confidence |
| BioMysteryBench (human-solvable) | 88.7% | estimated ± 0.6 pp, medium confidence |
| FrontierCode 1.1 Extended | 56.0% | estimated ± 3.4 pp, low confidence |
| HealthBench (raw) | 54.9% | estimated ± 6.0 pp, medium confidence |
| HealthBench (length-adjusted) | 53.8% | estimated ± 0.9 pp, medium confidence |
| HealthBench Professional | 56.8% | estimated ± 3.6 pp, high confidence |
| HealthBench Professional (raw) | 65.2% | estimated ± 4.5 pp, high confidence |
| HLE-Verified | 49.0% | estimated ± 1.3 pp, low confidence |
| LABBench2 | 83.6% | estimated ± 1.8 pp, low confidence |
| ScarfBench | 37.6% | measured |
| BlenderWorld (text-only) | 71.8% | estimated ± 2.8 pp, medium confidence |
| Cookie-Bench | 81.0% | estimated ± 1.3 pp, medium confidence |
| SimpleQA | 35.7% | estimated ± 7.8 pp, low confidence |
| Agents' Last Exam | 36.5% | estimated ± 11.0 pp, low confidence |
| BabyVision w/ Python | 95.2% | estimated ± 3.8 pp, low confidence |
| OpenHarmony Bench v1.0 | 58.9% | estimated ± 2.4 pp, high confidence |
| SEC-Bench Pro | 81.7% | estimated ± 5.6 pp, medium confidence |
| BabyVision | 73.6% | estimated ± 9.1 pp, low confidence |
| MMVU | 80.7% | estimated ± 2.7 pp, low confidence |
| PerceptionBench | 68.3% | estimated ± 2.7 pp, medium confidence |
| ZeroBench | 29.0% | measured |
| MRCRv2 | 85.9% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 61.5% | estimated ± 2.6 pp, low confidence |
| OCRBench V2 | 72.6% | estimated ± 3.3 pp, medium confidence |
| BioBasicsBench | 36.9% | estimated ± 2.8 pp, medium confidence |
| LifeVal | 75.2% | estimated ± 3.8 pp, medium confidence |
| MMMU-Pro w/ Python | 84.9% | estimated ± 2.0 pp, high confidence |
| ExploitBench Internal Port | 2.2% | estimated ± 1.4 pp, medium confidence |