Benchgen

MMLU-Pro — Results

RankModelScore
1namazu90.33
2qwen3-7-max89.6
3qwen3-6-plus88.5
4qwen3-7-plus88.5
5minimax-m2-188
6qwen3-5-397b-a17b87.8
7deepseek-v4-pro-max87.5
8kimi-k2-587.1
9ernie-5-087
10nemotron-3-ultra-550b-a55b86.8
11qwen3-5-122b-a10b86.7
12solar-pro-486.3
13deepseek-v4-flash-max86.2
14qwen3-6-27b86.2
15qwen3-5-27b86.1
16qwen3-5-35b-a3b85.3
17gemma-4-31b85.2
18qwen3-6-35b-a3b85.2
19deepseek-r1-052885
20deepseek-v3-2-exp85
21deepseek-v3-2-thinking85
22deepseek-v3-285
23mai-thinking-185
24mimo-v2-flash84.9
25glm-4-584.6

MMLU-Pro — Phases

1 phaseActive

Phase 1·

MMLU-Pro

Jun 19, 2026
Current