Mistral · model
Mistral Large 4 benchmark scores
As of 2026-11-22, Mistral Large 4 (Mistral) has measured scores on 26 benchmarks and estimated scores on 106 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 89.9% | estimated ± 9.3 pp, medium confidence |
| AA-HLE | 36.9% | estimated ± 5.8 pp, medium confidence |
| AA-SciCode | 56.6% | estimated ± 3.5 pp, medium confidence |
| Artificial Analysis Intelligence Index | 26.1% | measured |
| CritPt | 10.0% | measured |
| GDPval-AA | 46.1% | measured |
| GPQA Diamond | 90.7% | estimated ± 2.6 pp, high confidence |
| HLE | 31.3% | estimated ± 10.0 pp, low confidence |
| AA-Omniscience Accuracy | 26.0% | measured |
| Terminal-Bench 2.1 | 89.0% | estimated ± 5.4 pp, low confidence |
| BrowseComp | 84.5% | measured |
| HLE w/ tools | 51.3% | estimated ± 5.1 pp, medium confidence |
| VITA-Bench | 46.1% | estimated ± 6.7 pp, low confidence |
| τ²-bench results | 93.4% | estimated ± 11.8 pp, low confidence |
| AA Coding Index | 75.9% | estimated ± 2.3 pp, low confidence |
| APEX-Agents | 38.4% | estimated ± 9.4 pp, low confidence |
| APEX-Agents-AA | 32.5% | estimated ± 4.4 pp, high confidence |
| BioMysteryBench (human-difficult) | 43.8% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 88.2% | estimated ± 6.8 pp, low confidence |
| SWE-bench Verified | 87.3% | estimated ± 4.5 pp, medium confidence |
| AutomationBench | 47.1% | estimated ± 6.1 pp, low confidence |
| BFCL v4 | 76.3% | estimated ± 2.7 pp, medium confidence |
| CyberGym | 78.9% | estimated ± 6.0 pp, medium confidence |
| JobBench | 48.5% | estimated ± 8.7 pp, medium confidence |
| SkillsBench | 56.4% | estimated ± 4.1 pp, medium confidence |
| SWE-bench Pro | 59.1% | estimated ± 2.9 pp, low confidence |
| WideResearch | 76.5% | estimated ± 3.3 pp, high confidence |
| AA Agentic Index | 40.1% | estimated ± 3.1 pp, high confidence |
| MMLU-Pro | 85.1% | estimated ± 5.0 pp, medium confidence |
| FailureSensorIQ | 63.4% | estimated ± 1.7 pp, high confidence |
| GPQA | 89.6% | estimated ± 7.8 pp, medium confidence |
| MMLU | 90.5% | estimated ± 0.9 pp, high confidence |
| Vals GPQA Diamond | 87.7% | estimated ± 5.7 pp, medium confidence |
| Vals MMLU-Pro | 83.2% | estimated ± 2.9 pp, high confidence |
| MMMLU | 89.5% | estimated ± 0.7 pp, medium confidence |
| Chess Puzzles | 33.0% | measured |
| AA MMLU-Pro | 85.0% | estimated ± 0.8 pp, high confidence |
| Gert Labs | 39.8% | measured |
| AA-AnalystAgent | 42.1% | estimated ± 4.6 pp, high confidence |
| AA EnterpriseOps-Gym | 47.9% | estimated ± 7.2 pp, low confidence |
| AA Harvey LAB v1.0 | 91.6% | estimated ± 9.8 pp, low confidence |
| AA Tau3 Banking | 100.0% | estimated ± 2.4 pp, low confidence |
| ApprenticeBench | 4.2% | estimated ± 10.4 pp, low confidence |
| ARC-AGI-1 | 82.2% | estimated ± 7.8 pp, low confidence |
| ARC-AGI-2 | 49.8% | estimated ± 14.4 pp, low confidence |
| CursorBench 3.2 | 61.0% | estimated ± 4.1 pp, medium confidence |
| EBR-bench | 6.2% | estimated ± 9.3 pp, low confidence |
| Terminal-Bench 3.0 | 25.2% | estimated ± 3.2 pp, high confidence |
| FrontierCode 1.1 Main | 45.7% | estimated ± 3.9 pp, medium confidence |
| FrontierSWE v2 | 32.4% | estimated ± 8.0 pp, medium confidence |
| Furniture Assembly | 20.8% | measured |
| MaxxBench | 52.2% | estimated ± 5.7 pp, medium confidence |
| MazeBench | 0.2% | estimated ± 2.7 pp, high confidence |
| MirrorCode | 40.0% | measured |
| Mystery Game Puzzles | 17.0% | measured |
| NYT Connections | 27.4% | measured |
| OSWorld-Verified | 78.1% | estimated ± 5.7 pp, medium confidence |
| Terminal-Bench 4.1 | 27.6% | measured |
| Terminal-Bench Hard | 60.1% | estimated ± 2.8 pp, medium confidence |
| Vals SWE-bench | 80.1% | estimated ± 7.1 pp, low confidence |
| Terminal-Bench 2.1 (Vals AI) | 69.8% | estimated ± 6.9 pp, low confidence |
| CWE-Bench-AA | 58.3% | measured |
| AA Cyber Index | 50.9% | measured |
| DeepsecBench-AA | 15.0% | measured |
| Harvey LAB-AA v1.1 | 0.0% | estimated ± 3.5 pp, low confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 53.8% | estimated ± 4.3 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 55.1% | estimated ± 0.8 pp, low confidence |
| AA ITBench | 47.7% | estimated ± 4.1 pp, high confidence |
| Bug Hunt Bench | 26.1% | estimated ± 8.9 pp, medium confidence |
| CursorBench 4.0 | 46.7% | estimated ± 3.3 pp, medium confidence |
| CWE-bench v1 | 53.0% | estimated ± 5.2 pp, low confidence |
| DRACO | 83.0% | estimated ± 2.0 pp, medium confidence |
| FrontierCode 2.0 | 19.7% | measured |
| FrontierPhysics | 0.0% | measured |
| HLE w/o tools | 36.8% | estimated ± 4.9 pp, high confidence |
| OSWorld 2.0 | 60.3% | estimated ± 14.7 pp, low confidence |
| OSWorld 2.1 | 38.4% | estimated ± 3.6 pp, medium confidence |
| OSWorld 2.1 Offline | 48.1% | estimated ± 1.9 pp, low confidence |
| PostTrainBench v1.1 | 32.6% | estimated ± 4.8 pp, high confidence |
| SpireClimb | 0.0% | estimated ± 9.3 pp, medium confidence |
| StealthMark | 43.3% | measured |
| Terminal-Bench 4.0 | 30.9% | estimated ± 3.7 pp, medium confidence |
| Terminal-Bench-Science 0.1 | 5.3% | estimated ± 10.7 pp, low confidence |
| Toolathlon-Verified | 72.8% | estimated ± 0.8 pp, medium confidence |
| VulcanBench Frontier v4 | 82.9% | estimated ± 5.8 pp, low confidence |
| AA AutomationBench | 57.8% | measured |
| GDP.pdf | 17.6% | measured |
| AA Terminal-Bench 4.0 | 24.7% | measured |
| cursorBench41 | 40.2% | estimated ± 6.0 pp, low confidence |
| CWE-bench | 45.6% | measured |
| MKA | 35.1% | measured |
| SkillsBench official leaderboard (BenchFlow) | 44.9% | estimated ± 6.1 pp, medium confidence |
| BinaryAudit | 9.9% | estimated ± 12.1 pp, low confidence |
| CompileBench | 77.8% | estimated ± 10.0 pp, low confidence |
| OTelBench | 19.6% | estimated ± 4.1 pp, medium confidence |
| Vibe Code Bench | 78.4% | measured |
| FACTS Multimodal | 17.4% | estimated ± 4.0 pp, medium confidence |
| Hermes Bench | 71.4% | estimated ± 1.6 pp, medium confidence |
| Hermes Index | 37.4% | estimated ± 2.7 pp, medium confidence |
| ExploitBench | 32.4% | estimated ± 6.5 pp, low confidence |
| ExploitGym | 9.7% | estimated ± 7.2 pp, low confidence |
| C-Eval | 93.4% | estimated ± 1.0 pp, medium confidence |
| Claw-Eval | 77.4% | estimated ± 6.4 pp, medium confidence |
| MCP Atlas | 79.3% | estimated ± 10.3 pp, low confidence |
| MMLU-Redux | 94.2% | estimated ± 0.8 pp, medium confidence |
| PostTrainBench v1 | 29.8% | estimated ± 3.1 pp, low confidence |
| QwenClawBench | 93.2% | estimated ± 6.0 pp, low confidence |
| SuperGPQA | 66.1% | estimated ± 9.3 pp, medium confidence |
| Toolathlon | 54.3% | estimated ± 4.2 pp, high confidence |
| HealthBench Hard | 28.9% | estimated ± 8.6 pp, low confidence |
| React Native Evals | 85.4% | estimated ± 3.0 pp, low confidence |
| ResearchClawBench | 19.9% | estimated ± 2.2 pp, medium confidence |
| SWE-Rebench | 62.2% | estimated ± 6.5 pp, low confidence |
| PostTrainBench | 30.9% | estimated ± 4.5 pp, high confidence |
| DAYJOB | 1.4% | estimated ± 2.7 pp, medium confidence |
| DAYJOB Healthcare | 0.0% | estimated ± 4.0 pp, medium confidence |
| Finance Agent v2 | 54.7% | measured |
| BioMysteryBench (human-solvable) | 83.3% | estimated ± 0.6 pp, low confidence |
| HealthBench (length-adjusted) | 54.1% | estimated ± 3.7 pp, medium confidence |
| HealthBench Professional | 56.8% | estimated ± 4.2 pp, low confidence |
| HLE-Verified | 20.8% | estimated ± 7.1 pp, low confidence |
| ScarfBench | 67.7% | estimated ± 5.0 pp, low confidence |
| BlenderWorld (text-only) | 70.0% | estimated ± 2.8 pp, medium confidence |
| Cookie-Bench | 77.0% | estimated ± 1.3 pp, medium confidence |
| STEMBench | 2.5% | measured |
| SimpleQA | 26.3% | estimated ± 4.3 pp, medium confidence |
| Agents' Last Exam | 30.1% | estimated ± 9.7 pp, medium confidence |
| SEC-Bench Pro | 65.4% | estimated ± 6.9 pp, low confidence |
| BioBasicsBench | 34.1% | estimated ± 3.8 pp, low confidence |
| LifeVal | 75.7% | estimated ± 1.6 pp, medium confidence |
| ExploitBench Internal Port | 0.2% | estimated ± 1.4 pp, low confidence |
| PinchBench | 90.4% | estimated ± 3.6 pp, low confidence |