MiniMaxmodel systemcard
MiniMax M2.5
#25of 27 · general
superseded by M3, kept for the version ladder — measured by Vals, Epoch and Arena; AA and LiveBench list only M3
Standing
dot = this model · shaded = the field's middle 60% · track = full fieldGeneral
7.8#25/27
Math
9.6#25/27
Coding
34.8#16/27
Health
68.0#6/26
Legal
49.4#14/26
Instruct
0.0#27/27
Agents
9.7#25/27
Professional
0.0#27/27
Hallucination
0.0#27/27
Value
6.8#26/27
Measured
≈ imputed · amber dot = lab-self-reported among the sourcesGPQA Diamond82.1%#25
MMLU-Pro80.1%#23
AIME 202588.8%#6
SWE-bench V77.2%#20
LiveCodeBench79.2%#22
Terminal-Bench41.6%#26
APEX Agents6.2%#18
Arena1390#25
Epoch Capabilities Index146.7#21
Operations
1390arena elo · 40,783 votes