Kimi K2.5 vs Gemma 4 31B — Who Actually Wins? Real Benchmark Data (2026)

head to head

Kimi K2.5 vs Gemma 4 31B

10 shared benchmarks

10

wins

0

ties

0

wins

96.1%

AIME

89.2%

87.6%

GPQA Diamond

84.3%

24.37%

Humanity's Last Exam

19.5%

85.0%

LiveCodeBench

80.0%

87.1%

MMLU Pro

85.2%

78.5%

MMMU Pro

76.9%

73.0%

SWE-bench Multilingual

51.7%

50.7%

SWE-bench Pro

35.7%

76.8%

SWE-bench Verified

52.0%

50.8%

Terminal-Bench 2.0

42.9%