Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multimodal long-horizon complex reasoning on MM-Search
Loading...
71.3
Score
Claude-4.6-Sonnet-Thinking
33.652
43.426
53.2
62.974
Jun 13, 2026
Score
Updated 1mo ago
Evaluation Results
Method
Method
Links
Score
Claude-4.6-Sonnet-Thinking
Size=-, Train=-
2026.06
71.3
REDSearcher-MM-SFT
Size=30B, Train=Mid/SFT
2026.06
70.3
Vision-DeepResearch
Size=30B, Train=SFT/RL
2026.06
69.6
MM-DeepResearch
Size=32B, Train=SFT/RL
2026.06
69
Skywork-R1V4-30B
Size=30B, Train=SFT
2026.06
66.1
Gemini-3.1-Pro-Preview
Size=-, Train=-
2026.06
62
GLM-5
Size=744B, Train=-
2026.06
61.4
Kimi-K2.5
Size=1T, Train=-
2026.06
57.9
GPT-5.2
Size=-, Train=-
2026.06
56.7
MiniMax-M2.7
Size=230B, Train=-
2026.06
56.1
Qwen3.5-397B
Size=397B, Train=-
2026.06
55.6
S1-DeepResearch
Size=32B, Train=SFT
2026.06
54.4
Doubao-2.0-Pro
Size=-, Train=-
2026.06
50.9
DeepSeek-V3.2
Size=671B, Train=-
2026.06
46.8
Qwen3-235B
Size=235B, Train=-
2026.06
38.6
Qwen3-32B
Size=32B, Train=-
2026.06
35.1
Feedback
Search any
task
Search any
task