Kimi K2 Benchmark Details

Kimi K2 currently shows benchmark results led by MMLU (13 / 124, score 89.50), MATH-500 (12 / 46, score 97.40), Creative Writing (36 / 110, score 1665.70).

Benchmark Results

Kimi K2

Benchmark Results

Knowledge Exams

3 evaluations

Benchmark / mode

Score

Rank/total

MMLU

Standard Mode

89.50

13 / 124

MMLU-Pro

Standard Mode

81.10

58 / 175

HLE

Standard Mode

4.70

228 / 235

Abstract Generalization

1 evaluations

Benchmark / mode

Score

Rank/total

ARC-AGI-1

Standard Mode

13.30

182 / 193

Scientific Reasoning

1 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

Standard Mode

75.10

157 / 253

Honesty & Factuality

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleQA

Standard Mode

23 / 46

Repository Engineering

1 evaluations

Benchmark / mode

Score

Rank/total

SWE-bench Verified

Standard Mode

51.80

94 / 116

Mathematics

6 evaluations

Benchmark / mode

Score

Rank/total

MATH-500

Standard Mode

97.40

12 / 46

AIME 2024

Standard Mode

69.60

39 / 61

AIME2025

Standard Mode

94 / 110

IMO-ProofBench Advanced

Standard Mode

7.10

16 / 24

FrontierMath

Standard Mode

2.10

47 / 60

FrontierMath - Tier 4

Standard Mode

0.01

71 / 80

Algorithmic Coding

3 evaluations

Benchmark / mode

Score

Rank/total

LiveCodeBench

Standard Mode

53.70

90 / 125

IOI 2025 (Vals v1)

Standard ModeTools

7 / 9

IOI 2024 (Vals v1)

Standard ModeTools

0.50

10 / 10

Writing

1 evaluations

Benchmark / mode

Score

Rank/total

Creative Writing

Standard Mode

1665.70

36 / 110

Agentic Development

3 evaluations

Benchmark / mode

Score

Rank/total

Aider-Polyglot

Standard Mode

59.10

24 / 59

Terminal-Bench

Standard Mode

37.50

15 / 35

Terminal Bench Hard

Standard ModeTools

15.90

165 / 244

Commonsense

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleBench

Standard Mode

26.30

83 / 96

Service Workflows

3 evaluations

Benchmark / mode

Score

Rank/total

τ²-Bench - Telecom

Thinking ModeTools

65.80

24 / 30

τ²-Bench

Standard ModeTools

64.30

29 / 44

τ²-Bench

Thinking ModeTools

64.30

29 / 44

Cross-capability Suites

1 evaluations

Benchmark / mode

Score

Rank/total

LiveBench

Standard Mode

48.10

102 / 117

Compare with other models