Benchgen

MATH-500 — Results

RankModelScore
1minicpm5-2b94.6
2ling-3-0-flash-base79
3ling-3-0-tiny-base65.6
4longcat-flash-thinking0.992
5glm-4-50.982
6glm-4-5-air0.981
7kimi-k2-instruct-09050.974
8kimi-k2-instruct0.974
9llama-3-1-nemotron-ultra-253b-v10.97
10longcat-flash-lite0.968
11llama-3-3-nemotron-super-49b-v10.966
12longcat-flash-chat0.964
13kimi-k1-50.962
14deepseek-r1-05280.959
15deepseek-v3-03240.94
16qwq-32b0.906
17deepseek-v30.902

MATH-500 — Leaderboard

1 phaseActive

MATH-500 Leaderboard

No self-reported entries.