benchgap
MiniMax · model

MiniMax M3 benchmark scores

As of 2026-11-21, MiniMax M3 (MiniMax) has measured scores on 45 benchmarks and estimated scores on 145 more.

BenchmarkScoreSource
AA-GPQA Diamond92.9%measured
AA-HLE39.0%measured
AA-SciCode47.1%measured
Artificial Analysis Intelligence Index18.9%measured
CritPt3.7%measured
GDPval-AA37.1%measured
GPQA Diamond92.3%estimated ± 1.8 pp, high confidence
HLE43.4%estimated ± 8.5 pp, medium confidence
IFBench82.3%estimated ± 6.6 pp, low confidence
AA-LCR83.0%measured
LiveCodeBench v689.1%estimated ± 1.7 pp, high confidence
AA-Omniscience Accuracy44.9%measured
SciCode41.1%estimated ± 2.9 pp, low confidence
Terminal-Bench 2.166.0%measured
BrowseComp83.5%measured
HLE w/ tools51.7%estimated ± 5.3 pp, medium confidence
IFEval93.3%estimated ± 2.8 pp, medium confidence
LongBench v268.4%estimated ± 5.5 pp, medium confidence
VITA-Bench45.8%estimated ± 6.7 pp, low confidence
τ²-bench results88.9%measured
AA Coding Index58.6%measured
AA-MMMU-Pro78.6%measured
APEX-Agents29.0%estimated ± 9.4 pp, low confidence
APEX-Agents-AA28.2%estimated ± 2.9 pp, medium confidence
BioMysteryBench (human-difficult)33.6%estimated ± 6.8 pp, low confidence
DeepSearchQA89.4%estimated ± 4.5 pp, high confidence
SWE-bench Verified80.5%measured
AutomationBench47.6%estimated ± 9.5 pp, low confidence
BFCL v475.4%estimated ± 2.7 pp, medium confidence
CyberGym69.2%estimated ± 5.3 pp, medium confidence
JobBench27.7%estimated ± 4.9 pp, low confidence
SkillsBench55.3%estimated ± 4.1 pp, medium confidence
SWE-bench Pro59.0%measured
WideResearch81.3%estimated ± 2.1 pp, medium confidence
AA Agentic Index30.8%measured
MMLU-Pro84.4%estimated ± 4.8 pp, high confidence
AA-IFBench82.9%measured
FailureSensorIQ66.2%estimated ± 1.7 pp, high confidence
GPQA93.9%estimated ± 4.5 pp, high confidence
MMLU90.5%estimated ± 0.9 pp, medium confidence
Vals GPQA Diamond92.7%measured
Vals LiveCodeBench82.2%measured
Vals MMLU-Pro84.2%measured
MMMLU87.9%estimated ± 0.7 pp, medium confidence
MMMU85.2%estimated ± 0.8 pp, high confidence
Chess Puzzles17.0%measured
AA MMLU-Pro88.1%estimated ± 0.7 pp, medium confidence
Gert Labs56.9%estimated ± 4.6 pp, medium confidence
AA-AnalystAgent46.5%estimated ± 3.7 pp, low confidence
AA EnterpriseOps-Gym38.2%estimated ± 7.2 pp, low confidence
AA Harvey LAB v1.088.4%measured
AA Tau3 Banking47.1%estimated ± 2.4 pp, low confidence
AA Terminal-Bench 2.180.0%estimated ± 2.5 pp, medium confidence
ApprenticeBench0.0%estimated ± 6.3 pp, low confidence
ARC-AGI-190.9%estimated ± 6.5 pp, medium confidence
ARC-AGI-271.5%estimated ± 13.8 pp, low confidence
BabaIsBench46.0%measured
CursorBench 3.136.9%estimated ± 3.4 pp, low confidence
CursorBench 3.256.8%estimated ± 3.6 pp, high confidence
EBR-bench4.8%estimated ± 8.1 pp, low confidence
Terminal-Bench 3.012.1%estimated ± 3.2 pp, medium confidence
FrontierCode 1.1 Main28.4%estimated ± 4.2 pp, medium confidence
FrontierSWE v213.9%estimated ± 8.3 pp, medium confidence
Furniture Assembly32.9%estimated ± 9.3 pp, medium confidence
MaxxBench49.1%estimated ± 5.7 pp, medium confidence
MazeBench0.0%measured
Mystery Game Puzzles8.0%measured
NYT Connections65.1%measured
OfficeQA Pro45.1%measured
OSWorld-Verified70.1%measured
ParseBench-162.7%estimated ± 5.4 pp, low confidence
Terminal-Bench 4.17.6%measured
Terminal-Bench Hard54.6%estimated ± 2.8 pp, medium confidence
Vals SWE-bench75.0%measured
Terminal-Bench 2.1 (Vals AI)53.6%measured
VulcanBench v382.0%estimated ± 5.6 pp, low confidence
CWE-Bench-AA41.7%measured
AA Cyber Index22.4%measured
DeepsecBench-AA9.8%estimated ± 8.9 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)51.6%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)35.8%estimated ± 0.8 pp, low confidence
AA ITBench42.7%estimated ± 4.0 pp, high confidence
Bug Hunt Bench11.6%estimated ± 7.2 pp, low confidence
Chartography (tools)82.5%estimated ± 4.1 pp, low confidence
CursorBench 4.034.8%estimated ± 3.3 pp, medium confidence
CWE-bench v157.4%estimated ± 5.1 pp, medium confidence
DeepSWE29.9%estimated ± 6.4 pp, low confidence
DRACO86.9%estimated ± 0.7 pp, low confidence
FrontierCode 2.020.9%estimated ± 6.7 pp, medium confidence
FrontierPhysics1.7%estimated ± 4.7 pp, high confidence
HLE w/o tools28.5%estimated ± 4.9 pp, high confidence
OSWorld 2.04.6%measured
OSWorld 2.159.0%estimated ± 5.3 pp, low confidence
OSWorld 2.1 Offline20.7%estimated ± 1.9 pp, low confidence
PostTrainBench v1.119.8%estimated ± 3.7 pp, medium confidence
SpireClimb5.8%estimated ± 7.2 pp, low confidence
StealthMark38.6%measured
Terminal-Bench 4.015.8%estimated ± 4.5 pp, medium confidence
Toolathlon-Verified52.5%estimated ± 1.0 pp, low confidence
VulcanBench Frontier v477.3%estimated ± 6.8 pp, low confidence
AA AutomationBench48.5%estimated ± 8.2 pp, medium confidence
GDP.pdf17.0%estimated ± 4.6 pp, high confidence
AA Terminal-Bench 4.07.2%estimated ± 6.2 pp, medium confidence
ARC-AGI-30.0%estimated ± 8.6 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, medium confidence
cursorBench4144.8%estimated ± 3.7 pp, medium confidence
CWE-bench53.9%estimated ± 7.9 pp, low confidence
MKA41.9%estimated ± 5.7 pp, low confidence
SkillsBench official leaderboard (BenchFlow)53.0%measured
BinaryAudit37.6%estimated ± 12.0 pp, low confidence
CompileBench57.8%estimated ± 10.0 pp, low confidence
OTelBench22.5%estimated ± 6.6 pp, medium confidence
Vibe Code Bench52.2%estimated ± 8.4 pp, medium confidence
FACTS Multimodal37.3%estimated ± 4.0 pp, high confidence
Hermes Bench64.4%estimated ± 2.9 pp, high confidence
Hermes Index37.2%estimated ± 3.1 pp, high confidence
ParseBench-269.3%estimated ± 5.0 pp, medium confidence
CharXiv78.8%estimated ± 4.9 pp, high confidence
CharXiv w/o tools82.2%estimated ± 1.2 pp, high confidence
ExploitBench16.5%estimated ± 7.2 pp, low confidence
ExploitGym9.7%estimated ± 2.7 pp, medium confidence
AI2D_TEST91.8%estimated ± 1.3 pp, medium confidence
C-Eval91.2%estimated ± 1.0 pp, medium confidence
Claw-Eval74.5%measured
CountBench91.8%estimated ± 2.0 pp, medium confidence
DeepPlanning47.3%estimated ± 13.1 pp, low confidence
DynaMath84.7%estimated ± 2.1 pp, medium confidence
ERQA57.1%estimated ± 2.9 pp, high confidence
MathVision87.4%estimated ± 3.0 pp, high confidence
MCP Atlas74.2%measured
MLVU (M-Avg)86.0%estimated ± 2.2 pp, high confidence
MMLU-Redux96.6%estimated ± 0.8 pp, low confidence
MMMU-Pro78.1%measured
NL2Repo42.1%measured
OmniDocBench 1.591.6%measured
PostTrainBench v118.2%estimated ± 3.6 pp, high confidence
QwenClawBench61.9%estimated ± 2.6 pp, high confidence
RealWorldQA83.7%estimated ± 2.8 pp, high confidence
ScreenSpot Pro52.5%estimated ± 8.4 pp, low confidence
SimpleVQA72.9%estimated ± 7.1 pp, low confidence
SuperGPQA90.3%estimated ± 9.3 pp, medium confidence
Terminal-Bench 2.061.5%estimated ± 5.8 pp, medium confidence
Toolathlon53.2%estimated ± 4.2 pp, high confidence
V*100.0%estimated ± 7.5 pp, medium confidence
Video-MME (w/o subtitle)84.5%estimated ± 1.3 pp, high confidence
Video-MME (with subtitle)85.4%measured
VideoMMMU84.6%measured
We-Math74.5%estimated ± 3.5 pp, medium confidence
AA LiveCodeBench88.9%estimated ± 2.6 pp, high confidence
HealthBench Hard21.0%estimated ± 8.6 pp, low confidence
LiveCodeBench Pro83.0%estimated ± 6.0 pp, low confidence
MedXpertQA (MM)70.4%estimated ± 2.4 pp, high confidence
MMLU-Pro (Arcee)91.0%estimated ± 4.0 pp, low confidence
React Native Evals79.9%estimated ± 2.7 pp, medium confidence
ResearchClawBench19.8%measured
SWE-Rebench61.8%estimated ± 3.6 pp, medium confidence
PostTrainBench17.2%estimated ± 3.7 pp, medium confidence
DAYJOB0.0%estimated ± 3.2 pp, high confidence
DAYJOB Healthcare0.7%estimated ± 4.2 pp, high confidence
Finance Agent v256.1%estimated ± 3.8 pp, medium confidence
BioMysteryBench (human-solvable)81.7%estimated ± 0.6 pp, low confidence
FrontierCode 1.1 Extended55.7%estimated ± 3.0 pp, low confidence
HealthBench (raw)47.8%estimated ± 6.0 pp, medium confidence
HealthBench (length-adjusted)53.2%estimated ± 3.4 pp, medium confidence
HealthBench Professional56.7%estimated ± 4.2 pp, low confidence
HealthBench Professional (raw)58.4%estimated ± 4.5 pp, high confidence
HLE-Verified53.0%estimated ± 2.0 pp, medium confidence
LABBench279.9%estimated ± 1.8 pp, low confidence
ScarfBench25.8%estimated ± 9.1 pp, low confidence
BlenderWorld (text-only)70.0%estimated ± 2.8 pp, medium confidence
Cookie-Bench76.9%estimated ± 1.3 pp, medium confidence
STEMBench2.5%estimated ± 0.2 pp, medium confidence
SimpleQA52.3%estimated ± 4.3 pp, low confidence
LiveCodeBench79.6%estimated ± 14.6 pp, low confidence
Agents' Last Exam35.4%estimated ± 11.0 pp, low confidence
BabyVision w/ Python88.8%estimated ± 3.8 pp, low confidence
OpenHarmony Bench v1.048.4%measured
SEC-Bench Pro44.9%estimated ± 9.3 pp, medium confidence
BabyVision61.0%estimated ± 9.1 pp, low confidence
MMVU77.2%estimated ± 2.7 pp, medium confidence
PerceptionBench59.7%estimated ± 4.9 pp, low confidence
MRCRv285.9%estimated ± 11.3 pp, low confidence
CC-OCR82.5%estimated ± 3.7 pp, medium confidence
MMLongBench-Doc61.5%estimated ± 2.6 pp, low confidence
OCRBench V270.7%estimated ± 2.1 pp, medium confidence
BioBasicsBench33.9%estimated ± 2.8 pp, medium confidence
LifeVal74.9%estimated ± 1.6 pp, medium confidence
MMMU-Pro w/ Python81.5%estimated ± 2.0 pp, high confidence
ExploitBench Internal Port0.0%estimated ± 1.4 pp, low confidence
PinchBench90.4%estimated ± 3.6 pp, low confidence