Benchgen

MATH-500 — Results

RankModelScore
1twil-8b98.2
2lfm2-5-24b-a2b96.04
3minicpm5-2b94.6
4aliceai-foundation-80b-a3b-base91.1
5ling-3-0-flash-base79
6ling-3-0-tiny-base65.6
7longcat-flash-thinking0.992
8glm-4-50.982
9glm-4-5-air0.981
10kimi-k2-instruct-09050.974
11kimi-k2-instruct0.974
12llama-3-1-nemotron-ultra-253b-v10.97
13longcat-flash-lite0.968
14llama-3-3-nemotron-super-49b-v10.966
15longcat-flash-chat0.964
16kimi-k1-50.962
17deepseek-r1-05280.959
18deepseek-v3-03240.94
19qwq-32b0.906
20deepseek-v30.902

MATH-500 — Leaderboard

1 phaseActive

MATH-500 Leaderboard

No self-reported entries.