Benchgen

BrowseComp-zh — Results

RankModelScore
1minicpm5-2b43.5
2qwen3-5-397b-a17b0.703
3qwen3-5-122b-a10b0.699
4qwen3-5-35b-a3b0.695
5longcat-flash-thinking0.69
6glm-4-70.666
7deepseek-v3-2-thinking0.65
8deepseek-v3-20.65
9kimi-k2-thinking-09050.623
10qwen3-5-27b0.621
11deepseek-v3-10.492
12minimax-m20.485
13deepseek-v3-2-exp0.479
14deepseek-r1-05280.357
B

BrowseComp-zh — Phases

1 phaseActive

Phase 1·

BrowseComp-zh

Jul 24, 2026
Current