Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Mathematical Visual Question Answering on MATH-Vision Full (test)
Loading...
73.3
Relaxed Accuracy
Gemini 2.5 Pro
23.172
36.186
49.2
62.214
Sep 9, 2025
Relaxed Accuracy
Updated 3mo ago
Evaluation Results
Method
Method
Links
Relaxed Accuracy
Gemini 2.5 Pro
Model Category=Proprie...
2025.09
73.3
Human
Model Category=Baselin...
2025.09
68.82
Qwen2.5-VL-7B-Instruct + Visual-TableQA
Model Category=Finetun...
2025.09
49.77
Qwen2.5-VL-7B-Instruct + ReachQA
Model Category=Finetun...
2025.09
48.57
Llama 4 Maverick 17B-128E Instruct
Model Category=Open-So...
2025.09
45.89
Gemini 2.5 Flash
Model Category=Proprie...
2025.09
41.3
Qwen2.5-VL-32B-Instruct
Model Category=Open-So...
2025.09
38.1
Claude 3.5 Sonnet
Model Category=Proprie...
2025.09
32.76
Mistral Small 3.1 24B Instruct
Model Category=Open-So...
2025.09
32.45
GPT-4o
Model Category=Proprie...
2025.09
30.39
GPT-4o mini
Model Category=Proprie...
2025.09
28.85
Qwen2.5-VL-7B-Instruct
Model Category=Open-So...
2025.09
25.1
Feedback
Search any
task
Search any
task