evals.report
BenchmarksLabsCompareRun guidesIn the wild
Moonshot AIKimi

Kimi K2.6

Moonshot AI · Kimi. Released Apr 20, 2026.

Kimi K2.6 is a model from Moonshot AI in the Kimi family, released Apr 20, 2026. evals.report tracks 32 reported Kimi K2.6 benchmark scores across FrontierMath, DeepSWE, LiveBench, GPQA Diamond, SWE-bench Verified, Humanity's Last Exam, AIME (OTIS Mock), SimpleQA Verified, and 24 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

Open32 results

Benchmark results 32

Compare this model
BenchmarkCategoryScoreMetricStatusDate
FrontierMathReasoning38.97%accuracyOfficialApr 20, 2026Details
DeepSWECoding23.89%% resolvedOfficialApr 20, 2026Details
LiveBenchReasoning72.17%scoreOfficialApr 20, 2026Details
GPQA DiamondReasoning90.8%accuracyOfficialApr 20, 2026Details
SWE-bench VerifiedCoding76.7%% resolvedOfficialApr 20, 2026Details
Humanity's Last ExamReasoning29.9%accuracyOfficialApr 20, 2026Details
AIME (OTIS Mock)Reasoning96.1%accuracyOfficialApr 20, 2026Details
SimpleQA VerifiedOther38.7%accuracyOfficialApr 20, 2026Details
LMArenaChat preference1456source-defined ratingOfficialApr 20, 2026Details
GBA EvalCoding0.9%overall scoreOfficialApr 20, 2026Details
WeirdMLCoding55.9%average accuracyOfficialApr 20, 2026Details
CursorBenchAgents47.6%scoreOfficialApr 20, 2026Details
Artificial Analysis Intelligence IndexReasoning53.9IndexUnverifiedApr 20, 2026Details
Epoch Capabilities IndexReasoning151.6IndexOfficialApr 20, 2026Details
OSWorldAgents73.1%task success rateUnverifiedApr 20, 2026Details
τ²-bench (Telecom)Tool use95.9%pass^1OfficialApr 20, 2026Details
AIME 2026Reasoning95.83%accuracyOfficialApr 20, 2026Details
GDPvalAgents1481EloOfficialApr 20, 2026Details
SciCodeCoding53.5%accuracyUnverifiedApr 20, 2026Details
AA-Omniscience: Knowledge and Hallucination BenchmarkReasoning6AA-Omniscience IndexOfficialApr 20, 2026Details
IFBenchReasoning76.0%accuracyOfficialApr 20, 2026Details
WebDev ArenaChat preference1518EloVerifiedApr 20, 2026Details
EQ-Bench Creative Writing v3Chat preference1782EloVerifiedApr 20, 2026Details
Design ArenaChat preference1335EloVerifiedApr 20, 2026Details
CharXivMultimodal86.7%accuracyUnverifiedApr 20, 2026Details
MathArena HMMT February 2026Reasoning94.70%accuracyOfficialApr 20, 2026Details
FrontierMath Tier 4Reasoning14.6%accuracyOfficialApr 20, 2026Details
Vectara Hallucination LeaderboardOther10.8%Hallucination RateOfficialApr 20, 2026Details
Vibe Code BenchCoding37.89%Overall accuracyVerifiedApr 20, 2026Details
Terminal-Bench 2.0Agents66.7%task successVerifiedApr 20, 2026Details
FrontierSWEAgents27%dominance scoreOfficialApr 20, 2026Details
FrontierCodeCoding3.8%weighted score (Diamond)OfficialApr 20, 2026Details