evals.report
BenchmarksLabsCompareRun guidesIn the wild
Moonshot AIKimi

Kimi K2.5

Moonshot AI · Kimi. Released Jan 27, 2026.

Kimi K2.5 is a model from Moonshot AI in the Kimi family, released Jan 27, 2026. evals.report tracks 35 reported Kimi K2.5 benchmark scores across FrontierMath, ARC-AGI-1, GPQA Diamond, SWE-bench Verified, AIME (OTIS Mock), SimpleQA Verified, ARC-AGI-2, WeirdML, and 27 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

Open35 results

Benchmark results 35

Compare this model
BenchmarkCategoryScoreMetricStatusDate
FrontierMathReasoning27.9%accuracyOfficialJan 27, 2026Details
ARC-AGI-1Reasoning65.33%accuracyOfficialJan 27, 2026Details
GPQA DiamondReasoning87.6%accuracyOfficialJan 27, 2026Details
SWE-bench VerifiedCoding73.8%% resolvedOfficialJan 27, 2026Details
AIME (OTIS Mock)Reasoning92.2%accuracyOfficialJan 27, 2026Details
SimpleQA VerifiedOther33.9%accuracyOfficialJan 27, 2026Details
ARC-AGI-2Reasoning11.81%accuracyOfficialJan 27, 2026Details
WeirdMLCoding45.6%average accuracyOfficialJan 27, 2026Details
CursorBenchAgents31.9%scoreOfficialJan 27, 2026Details
PostTrainBenchAgents10.26%weighted average scoreOfficialJan 27, 2026Details
Artificial Analysis Intelligence IndexReasoning46.8IndexUnverifiedJan 27, 2026Details
Epoch Capabilities IndexReasoning148.2IndexOfficialJan 27, 2026Details
SWE-rebenchCoding58.5%Resolved rate (pass@1)UnverifiedJan 27, 2026Details
GAIA: A Benchmark for General AI AssistantsAgents38.1%accuracyUnverifiedJan 27, 2026Details
AIME 2026Reasoning95.83%accuracyOfficialJan 27, 2026Details
Video-MMEMultimodal87.4%accuracyUnverifiedJan 27, 2026Details
GDPvalAgents1285EloOfficialJan 27, 2026Details
SciCodeCoding49.0%accuracyUnverifiedJan 27, 2026Details
MultiChallengeReasoning61.39%accuracyVerifiedJan 27, 2026Details
Global-MMLUReasoning84.0%accuracyUnverifiedJan 27, 2026Details
Video-MMMUMultimodal86.6%accuracyUnverifiedJan 27, 2026Details
WebDev ArenaChat preference1431EloVerifiedJan 27, 2026Details
EQ-Bench Creative Writing v3Chat preference1593EloVerifiedJan 27, 2026Details
Design ArenaChat preference1292EloVerifiedJan 27, 2026Details
MASK (Model Alignment between Statements and Knowledge)Other70.47Honesty scoreVerifiedJan 27, 2026Details
CharXivMultimodal77.5%accuracyUnverifiedJan 27, 2026Details
SWE-bench MultilingualCoding67.3%% resolvedOfficialJan 27, 2026Details
EnigmaEvalReasoning3.38%accuracyVerifiedJan 27, 2026Details
MathArena HMMT February 2026Reasoning87.12%accuracyOfficialJan 27, 2026Details
FrontierMath Tier 4Reasoning4.2%accuracyOfficialJan 27, 2026Details
Vectara Hallucination LeaderboardOther14.2%Hallucination RateOfficialJan 27, 2026Details
Vibe Code BenchCoding17.54%Overall accuracyVerifiedJan 27, 2026Details
MultiNRCReasoning35.17%accuracyOfficialJan 27, 2026Details
FrontierSWEAgents26%dominance scoreOfficialJan 27, 2026Details
FrontierCodeCoding1.0%weighted score (Diamond)OfficialJan 27, 2026Details