benchgap
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.

BenchmarkScoreSource
AA-GPQA Diamond89.9%estimated ± 9.3 pp, medium confidence
AA-HLE36.9%estimated ± 5.8 pp, medium confidence
AA-SciCode56.6%estimated ± 3.5 pp, medium confidence
Artificial Analysis Intelligence Index26.1%measured
CritPt10.0%measured
GDPval-AA46.1%measured
GPQA Diamond90.7%estimated ± 2.6 pp, high confidence
HLE31.3%estimated ± 10.0 pp, low confidence
AA-Omniscience Accuracy26.0%measured
Terminal-Bench 2.189.0%estimated ± 5.4 pp, low confidence
BrowseComp84.5%measured
HLE w/ tools51.3%estimated ± 5.1 pp, medium confidence
VITA-Bench46.1%estimated ± 6.7 pp, low confidence
τ²-bench results93.4%estimated ± 11.8 pp, low confidence
AA Coding Index75.9%estimated ± 2.3 pp, low confidence
APEX-Agents38.4%estimated ± 9.4 pp, low confidence
APEX-Agents-AA32.5%estimated ± 4.4 pp, high confidence
BioMysteryBench (human-difficult)43.8%estimated ± 6.8 pp, low confidence
DeepSearchQA88.2%estimated ± 6.8 pp, low confidence
SWE-bench Verified87.3%estimated ± 4.5 pp, medium confidence
AutomationBench47.1%estimated ± 6.1 pp, low confidence
BFCL v476.3%estimated ± 2.7 pp, medium confidence
CyberGym78.9%estimated ± 6.0 pp, medium confidence
JobBench48.5%estimated ± 8.7 pp, medium confidence
SkillsBench56.4%estimated ± 4.1 pp, medium confidence
SWE-bench Pro59.1%estimated ± 2.9 pp, low confidence
WideResearch76.5%estimated ± 3.3 pp, high confidence
AA Agentic Index40.1%estimated ± 3.1 pp, high confidence
MMLU-Pro85.1%estimated ± 5.0 pp, medium confidence
FailureSensorIQ63.4%estimated ± 1.7 pp, high confidence
GPQA89.6%estimated ± 7.8 pp, medium confidence
MMLU90.5%estimated ± 0.9 pp, high confidence
Vals GPQA Diamond87.7%estimated ± 5.7 pp, medium confidence
Vals MMLU-Pro83.2%estimated ± 2.9 pp, high confidence
MMMLU89.5%estimated ± 0.7 pp, medium confidence
Chess Puzzles33.0%measured
AA MMLU-Pro85.0%estimated ± 0.8 pp, high confidence
Gert Labs39.8%measured
AA-AnalystAgent42.1%estimated ± 4.6 pp, high confidence
AA EnterpriseOps-Gym47.9%estimated ± 7.2 pp, low confidence
AA Harvey LAB v1.091.6%estimated ± 9.8 pp, low confidence
AA Tau3 Banking100.0%estimated ± 2.4 pp, low confidence
ApprenticeBench4.2%estimated ± 10.4 pp, low confidence
ARC-AGI-182.2%estimated ± 7.8 pp, low confidence
ARC-AGI-249.8%estimated ± 14.4 pp, low confidence
CursorBench 3.261.0%estimated ± 4.1 pp, medium confidence
EBR-bench6.2%estimated ± 9.3 pp, low confidence
Terminal-Bench 3.025.2%estimated ± 3.2 pp, high confidence
FrontierCode 1.1 Main45.7%estimated ± 3.9 pp, medium confidence
FrontierSWE v232.4%estimated ± 8.0 pp, medium confidence
Furniture Assembly20.8%measured
MaxxBench52.2%estimated ± 5.7 pp, medium confidence
MazeBench0.2%estimated ± 2.7 pp, high confidence
MirrorCode40.0%measured
Mystery Game Puzzles17.0%measured
NYT Connections27.4%measured
OSWorld-Verified78.1%estimated ± 5.7 pp, medium confidence
Terminal-Bench 4.127.6%measured
Terminal-Bench Hard60.1%estimated ± 2.8 pp, medium confidence
Vals SWE-bench80.1%estimated ± 7.1 pp, low confidence
Terminal-Bench 2.1 (Vals AI)69.8%estimated ± 6.9 pp, low confidence
CWE-Bench-AA58.3%measured
AA Cyber Index50.9%measured
DeepsecBench-AA15.0%measured
Harvey LAB-AA v1.10.0%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 ITBench47.7%estimated ± 4.1 pp, high confidence
Bug Hunt Bench26.1%estimated ± 8.9 pp, medium confidence
CursorBench 4.046.7%estimated ± 3.3 pp, medium confidence
CWE-bench v153.0%estimated ± 5.2 pp, low confidence
DRACO83.0%estimated ± 2.0 pp, medium confidence
FrontierCode 2.019.7%measured
FrontierPhysics0.0%measured
HLE w/o tools36.8%estimated ± 4.9 pp, high confidence
OSWorld 2.060.3%estimated ± 14.7 pp, low confidence
OSWorld 2.138.4%estimated ± 3.6 pp, medium confidence
OSWorld 2.1 Offline48.1%estimated ± 1.9 pp, low confidence
PostTrainBench v1.132.6%estimated ± 4.8 pp, high confidence
SpireClimb0.0%estimated ± 9.3 pp, medium confidence
StealthMark43.3%measured
Terminal-Bench 4.030.9%estimated ± 3.7 pp, medium confidence
Terminal-Bench-Science 0.15.3%estimated ± 10.7 pp, low confidence
Toolathlon-Verified72.8%estimated ± 0.8 pp, medium confidence
VulcanBench Frontier v482.9%estimated ± 5.8 pp, low confidence
AA AutomationBench57.8%measured
GDP.pdf17.6%measured
AA Terminal-Bench 4.024.7%measured
cursorBench4140.2%estimated ± 6.0 pp, low confidence
CWE-bench45.6%measured
MKA35.1%measured
SkillsBench official leaderboard (BenchFlow)44.9%estimated ± 6.1 pp, medium confidence
BinaryAudit9.9%estimated ± 12.1 pp, low confidence
CompileBench77.8%estimated ± 10.0 pp, low confidence
OTelBench19.6%estimated ± 4.1 pp, medium confidence
Vibe Code Bench78.4%measured
FACTS Multimodal17.4%estimated ± 4.0 pp, medium confidence
Hermes Bench71.4%estimated ± 1.6 pp, medium confidence
Hermes Index37.4%estimated ± 2.7 pp, medium confidence
ExploitBench32.4%estimated ± 6.5 pp, low confidence
ExploitGym9.7%estimated ± 7.2 pp, low confidence
C-Eval93.4%estimated ± 1.0 pp, medium confidence
Claw-Eval77.4%estimated ± 6.4 pp, medium confidence
MCP Atlas79.3%estimated ± 10.3 pp, low confidence
MMLU-Redux94.2%estimated ± 0.8 pp, medium confidence
PostTrainBench v129.8%estimated ± 3.1 pp, low confidence
QwenClawBench93.2%estimated ± 6.0 pp, low confidence
SuperGPQA66.1%estimated ± 9.3 pp, medium confidence
Toolathlon54.3%estimated ± 4.2 pp, high confidence
HealthBench Hard28.9%estimated ± 8.6 pp, low confidence
React Native Evals85.4%estimated ± 3.0 pp, low confidence
ResearchClawBench19.9%estimated ± 2.2 pp, medium confidence
SWE-Rebench62.2%estimated ± 6.5 pp, low confidence
PostTrainBench30.9%estimated ± 4.5 pp, high confidence
DAYJOB1.4%estimated ± 2.7 pp, medium confidence
DAYJOB Healthcare0.0%estimated ± 4.0 pp, medium confidence
Finance Agent v254.7%measured
BioMysteryBench (human-solvable)83.3%estimated ± 0.6 pp, low confidence
HealthBench (length-adjusted)54.1%estimated ± 3.7 pp, medium confidence
HealthBench Professional56.8%estimated ± 4.2 pp, low confidence
HLE-Verified20.8%estimated ± 7.1 pp, low confidence
ScarfBench67.7%estimated ± 5.0 pp, low confidence
BlenderWorld (text-only)70.0%estimated ± 2.8 pp, medium confidence
Cookie-Bench77.0%estimated ± 1.3 pp, medium confidence
STEMBench2.5%measured
SimpleQA26.3%estimated ± 4.3 pp, medium confidence
Agents' Last Exam30.1%estimated ± 9.7 pp, medium confidence
SEC-Bench Pro65.4%estimated ± 6.9 pp, low confidence
BioBasicsBench34.1%estimated ± 3.8 pp, low confidence
LifeVal75.7%estimated ± 1.6 pp, medium confidence
ExploitBench Internal Port0.2%estimated ± 1.4 pp, low confidence
PinchBench90.4%estimated ± 3.6 pp, low confidence