Benchgen

RealWorldQA — Results

RankModelScore
1qwen3-8-flash-next88.5
2qwen3-8-27b85.9
3lfm3-vl-2b79.3
4lfm2-5-vl-3b73.1
5lfm3-omni-2b64.68
6north-micro-vision-instruct62.2
7qwen3-7-plus0.869
8seed-2-1-pro0.867
9seed-2-1-turbo0.863
10qwen3-6-plus0.854
11qwen3-6-35b-a3b0.853
12qwen3-5-122b-a10b0.851
13qwen3-5-35b-a3b0.841
14qwen3-6-27b0.841
15qwen3-5-27b0.837
16qwen3-vl-235b-a22b-thinking0.813
17qwen3-vl-235b-a22b-instruct0.793
18qwen3-vl-32b-instruct0.79
19qwen3-vl-32b-thinking0.784
20qwen3-vl-30b-a3b-thinking0.774
21qwen3-vl-30b-a3b-instruct0.737
22qwen3-vl-8b-thinking0.735
23qwen3-vl-4b-thinking0.732
24qwen3-vl-8b-instruct0.715
25qwen3-vl-4b-instruct0.709

RealWorldQA — Phases

1 phaseActive

Phase 1·

RealWorldQA

Aug 3, 2026
Current