benchgap
Moonshot · model

Kimi K2.5 benchmark scores

As of 2026-11-21, Kimi K2.5 (Moonshot) has measured scores on 77 benchmarks and estimated scores on 102 more.

BenchmarkScoreSource
AA-GPQA Diamond87.9%measured
AA-HLE30.7%measured
AA-SciCode48.2%estimated ± 3.0 pp, high confidence
AIME2695.8%measured
Artificial Analysis Intelligence Index23.5%measured
CritPt3.1%measured
GDPval-AA17.0%measured
GPQA Diamond87.6%measured
HLE30.1%measured
IFBench68.2%estimated ± 6.6 pp, medium confidence
AA-LCR78.0%measured
LiveCodeBench v685.0%measured
AA-Omniscience Accuracy35.2%measured
SciCode48.7%measured
BrowseComp60.6%measured
HLE w/ tools38.6%estimated ± 4.1 pp, low confidence
IFEval93.9%measured
LongBench v261.0%measured
VITA-Bench32.2%estimated ± 6.7 pp, low confidence
τ²-bench results95.9%measured
AA Coding Index46.8%measured
AA-MMMU-Pro75.4%measured
APEX-Agents11.7%estimated ± 9.4 pp, low confidence
APEX-Agents-AA11.5%measured
BioMysteryBench (human-difficult)31.4%estimated ± 6.8 pp, low confidence
DeepSearchQA77.1%measured
SWE-bench Verified76.8%measured
AutomationBench48.3%estimated ± 9.1 pp, low confidence
BFCL v468.3%measured
CyberGym39.0%estimated ± 6.0 pp, low confidence
JobBench8.7%measured
SkillsBench33.5%estimated ± 4.1 pp, low confidence
SWE-bench Pro50.7%measured
τ³-bench results65.7%measured
WideResearch72.7%measured
AA Agentic Index12.5%estimated ± 3.1 pp, high confidence
MMLU-Pro87.1%measured
AA-IFBench70.2%measured
AIME 202495.0%estimated ± 2.6 pp, high confidence
BBH100.0%estimated ± 2.8 pp, medium confidence
FailureSensorIQ66.2%estimated ± 1.8 pp, medium confidence
GPQA87.6%measured
MMLU90.5%estimated ± 0.9 pp, medium confidence
Vals GPQA Diamond86.9%estimated ± 2.1 pp, high confidence
Vals LiveCodeBench86.0%estimated ± 0.8 pp, medium confidence
Vals MMLU-Pro85.3%estimated ± 1.6 pp, high confidence
MATH-50099.0%estimated ± 2.0 pp, high confidence
MMMLU87.7%estimated ± 0.7 pp, medium confidence
MMMU84.3%measured
AIME 202596.1%measured
AA MMLU-Pro86.4%estimated ± 0.8 pp, high confidence
Gert Labs45.9%measured
AA EnterpriseOps-Gym35.0%estimated ± 1.2 pp, low confidence
AA Harvey LAB v1.085.6%estimated ± 9.8 pp, low confidence
AA Tau3 Banking39.2%estimated ± 2.7 pp, low confidence
ApprenticeBench0.0%estimated ± 6.0 pp, low confidence
ARC-AGI-185.5%estimated ± 2.5 pp, medium confidence
ARC-AGI-243.9%estimated ± 9.7 pp, low confidence
CursorBench 3.148.2%estimated ± 5.6 pp, low confidence
CursorBench 3.241.8%estimated ± 3.6 pp, medium confidence
FrontierCode 1.1 Main21.8%estimated ± 4.2 pp, low confidence
FrontierMath v2 (Tier 4)10.7%estimated ± 4.1 pp, medium confidence
FrontierMath v2 (Tiers 1-3)33.0%estimated ± 7.8 pp, low confidence
FrontierSWE v20.0%estimated ± 8.3 pp, low confidence
MazeBench0.0%estimated ± 5.1 pp, low confidence
OfficeQA Pro50.3%estimated ± 6.8 pp, low confidence
OSWorld-Verified63.3%measured
ParseBench-159.6%estimated ± 6.2 pp, medium confidence
Terminal-Bench Hard42.4%estimated ± 4.0 pp, medium confidence
Vals SWE-bench65.7%estimated ± 5.2 pp, low confidence
Terminal-Bench 2.1 (Vals AI)44.5%estimated ± 6.4 pp, medium confidence
CWE-Bench-AA40.9%estimated ± 6.4 pp, medium confidence
AA Cyber Index30.6%estimated ± 13.5 pp, low confidence
DeepsecBench-AA0.1%estimated ± 8.9 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)53.0%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)32.3%estimated ± 0.8 pp, low confidence
AA ITBench36.9%estimated ± 2.8 pp, high confidence
Bug Hunt Bench7.3%estimated ± 7.2 pp, low confidence
Chartography (tools)80.5%estimated ± 4.1 pp, low confidence
CursorBench 4.028.6%estimated ± 3.3 pp, medium confidence
CWE-bench v125.3%estimated ± 6.4 pp, low confidence
DeepSWE47.6%estimated ± 8.4 pp, low confidence
DRACO64.2%estimated ± 2.0 pp, low confidence
FrontierCode 2.019.1%estimated ± 6.7 pp, low confidence
HLE w/o tools33.3%estimated ± 2.7 pp, high confidence
OSWorld 2.00.0%estimated ± 3.6 pp, medium confidence
PostTrainBench v1.120.3%estimated ± 4.4 pp, medium confidence
SWE Multilingual73.0%measured
Toolathlon-Verified40.3%estimated ± 3.1 pp, medium confidence
VulcanBench Frontier v474.0%estimated ± 8.3 pp, low confidence
GDP.pdf16.4%estimated ± 4.1 pp, low confidence
ARC-AGI-30.0%estimated ± 8.3 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, low confidence
SkillsBench official leaderboard (BenchFlow)46.2%estimated ± 9.5 pp, low confidence
BinaryAudit24.6%measured
OTelBench19.9%estimated ± 5.6 pp, medium confidence
Vibe Code Bench27.9%estimated ± 6.5 pp, low confidence
Hermes Bench46.7%estimated ± 2.9 pp, medium confidence
Hermes Index16.2%estimated ± 8.7 pp, low confidence
ParseBench-269.8%estimated ± 5.0 pp, medium confidence
CharXiv77.5%measured
CharXiv w/o tools82.6%estimated ± 1.2 pp, high confidence
ExploitGym0.1%estimated ± 10.6 pp, low confidence
AI2D_TEST90.8%measured
C-Eval94.0%measured
Claw-Eval52.3%measured
CountBench94.1%measured
DeepPlanning14.4%measured
DynaMath84.4%measured
ERQA65.8%estimated ± 2.7 pp, medium confidence
HMMT Feb 202595.4%measured
HMMT Feb 202687.1%measured
HMMT Nov 202591.1%measured
INCLUDE83.3%measured
MathVision84.2%measured
MAXIFE72.8%measured
MCP Atlas29.5%measured
MCP-Tasks59.1%measured
MLVU (M-Avg)85.0%measured
MMAnswerBench81.8%measured
MMLU-ProX82.3%measured
MMLU-Redux94.5%measured
MMMU-Pro78.5%measured
NL2Repo32.0%measured
NOVA-6356.0%measured
OmniDocBench 1.588.8%measured
PolyMath43.1%measured
PostTrainBench v115.5%estimated ± 4.6 pp, medium confidence
QwenClawBench54.3%measured
RealWorldQA81.0%measured
ScreenSpot Pro61.7%estimated ± 4.8 pp, medium confidence
SimpleVQA71.2%measured
SuperGPQA69.2%measured
Terminal-Bench 2.050.8%measured
Toolathlon27.8%measured
V*99.0%estimated ± 5.8 pp, low confidence
Video-MME (w/o subtitle)83.2%measured
Video-MME (with subtitle)87.4%measured
VideoMMMU86.6%measured
We-Math84.7%measured
AA LiveCodeBench88.1%estimated ± 4.7 pp, low confidence
HealthBench Hard0.0%estimated ± 5.1 pp, low confidence
LiveCodeBench Pro76.0%estimated ± 6.0 pp, low confidence
MedXpertQA (MM)71.3%estimated ± 2.4 pp, high confidence
MedXpertQA (Text)47.4%estimated ± 4.5 pp, low confidence
MMLU-Pro (Arcee)87.1%measured
React Native Evals77.2%measured
ResearchClawBench14.0%measured
SWE-Rebench58.5%measured
PostTrainBench18.8%estimated ± 3.7 pp, medium confidence
BioMysteryBench (human-solvable)81.7%estimated ± 0.6 pp, low confidence
FrontierCode 1.1 Extended52.6%estimated ± 3.0 pp, low confidence
HealthBench (raw)39.6%estimated ± 6.0 pp, low confidence
HealthBench (length-adjusted)50.0%estimated ± 3.4 pp, low confidence
HealthBench Professional56.7%estimated ± 4.2 pp, low confidence
HealthBench Professional (raw)50.8%estimated ± 4.5 pp, medium confidence
HLE-Verified15.4%estimated ± 2.0 pp, low confidence
LABBench273.8%estimated ± 1.8 pp, low confidence
ScarfBench24.5%estimated ± 9.1 pp, low confidence
STEMBench2.5%estimated ± 0.2 pp, medium confidence
SimpleQA39.0%estimated ± 4.3 pp, low confidence
LiveCodeBench77.3%estimated ± 14.6 pp, low confidence
Agents' Last Exam14.0%estimated ± 11.6 pp, low confidence
BabyVision w/ Python88.8%estimated ± 3.8 pp, low confidence
OpenHarmony Bench v1.029.0%estimated ± 2.7 pp, medium confidence
IMOAnswerBench81.6%estimated ± 5.6 pp, medium confidence
BabyVision43.1%estimated ± 3.7 pp, high confidence
MMVU80.4%measured
PerceptionBench50.9%estimated ± 3.6 pp, low confidence
ZeroBench15.7%estimated ± 3.6 pp, low confidence
MRCRv285.5%estimated ± 11.3 pp, low confidence
CC-OCR79.7%measured
MMLongBench-Doc58.5%measured
RefCOCO (avg)87.8%measured
OCRBench V271.9%estimated ± 2.1 pp, medium confidence
BioBasicsBench33.9%estimated ± 2.8 pp, low confidence
LifeVal38.4%estimated ± 1.6 pp, low confidence
MMMU-Pro w/ Python78.6%estimated ± 2.0 pp, high confidence
PinchBench88.9%estimated ± 3.6 pp, medium confidence