#ModelAccuracyPaper
1Qwen/QwQ-32B90.60link
2OpenThinker3-7B90.00link
3deepseek-reasoner89.60link
4Qwen-2.5-7B-Instruct89.00link
5OpenThinker-7B84.40link
6microsoft/Phi-4-reasoning-plus84.00link
7Llama-3.1-8B-Instruct83.80link
8w/ R1 template83.80link
9FD + AD40.60link
10FD40.40link
11EB40.00link
12EB + AD40.00link
13Top-k28.40link
14Top-k + AD28.40link
MATH500 math500 Leaderboard