Share your thoughts, 1 month free Claude Pro on usSee more

STEM Reasoning on MathVision

69.96Accuracy

CodePercept-32B-S1

Updated 1mo ago

Evaluation Results

Method	Links
CodePercept-32B-S1 2026.03		69.96
Gemini2.5-Pro 2026.03		66.8
CodePercept-8B-S1 2026.03		66.45
CodePercept-4B-S1 2026.03		64.71
Qwen3-VL-32B-Instruct 2026.03		62.66
CodePercept-32B-S1 2026.03		62.27
Qwen3-VL-235A22B-Instruct 2026.03		60.43
GPT5-Thinking 2026.03		60.03
Qwen3-VL-4B-Instruct 2026.03		59.8
Qwen3-VL-8B-Instruct 2026.03		59.67
Claude-Opus 4.1-Thinking 2026.03		59.61
CodePercept-8B-S1 2026.03		59.31
Qwen3-VL-32B-Instruct 2026.03		58.55
CodePercept-4B-S1 2026.03		57.63
Qwen3-VL-8B-Instruct 2026.03		54.37
Qwen3-VL-4B-Instruct 2026.03		54.21
Qwen2.5-VL-72B 2026.03		54.14
KeyeVL1.5-8B 2026.03		54.11
GLM-4.1V-9B 2026.03		53.75
Qwen3-VL-30A3B-Instruct 2026.03		53.59
InternVL3.5-8B 2026.03		53.32
MiniCPM-V-4.5 2026.03		53.15
Intern-S1-8B 2026.03		51.67