Kimi K2.6 — Benchmarks

Benchmark scores for Kimi K2.6 aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #29 of 78 open modelscomposite 51.7/100 across 15 benchmarks in 4 categories · methodology

Coding

BenchmarkScoreOpen rankAll models
LiveBench Coding78.6#6 / 19#27 / 63
SciCode53.5%#3 / 62#48 / 161
WeirdML55.9%#7 / 46#57 / 161
LMArena WebDev1508.6#12 / 48#43 / 123

Instruction Following

BenchmarkScoreOpen rankAll models
LMArena Text1460.4#8 / 182#52 / 391

Knowledge

BenchmarkScoreOpen rankAll models
SimpleQA34.9%#8 / 15#45 / 80

Math

BenchmarkScoreOpen rankAll models
FrontierMath57.2%#6 / 15#35 / 102
FrontierMath Tier 425.6%#5 / 11#38 / 64
ProofBench16.0%#13 / 24#45 / 68
LiveBench Math84.3#14 / 19#52 / 63
AIME 2024/202596.1%#4 / 82#29 / 270

Reasoning

BenchmarkScoreOpen rankAll models
Chess Puzzles26.0%#4 / 65#60 / 205
Mystery Game Puzzles18.0%#8 / 17#60 / 122
DTBench90.9%#5 / 67#60 / 210
LMCA37.3%#12 / 53#96 / 172
GPQA Diamond90.8%#8 / 94#33 / 289
CritPt8.0%#11 / 62#78 / 172
LiveBench Reasoning79.4#12 / 19#49 / 63

Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.