Our new X account is live! Follow @wizwand_team for updates

Multimodal Reasoning on MATH-Vision (full)

62.7Accuracy

Qwen3-VL-8B-Thinking

Updated 3d ago

Evaluation Results

Method	Links
Qwen3-VL-8B-Thinking 2025.12		62.7
Qwen3-VL-8B-Instruct + CARE 2025.12		61.7
MiMo-VL-7B-RL 2025.12		60.4
Qwen3-VL-4B-Thinking 2025.12		60
GPT-5-Nano 2025.12		59.7
MiMo-VL-7B-SFT 2025.12		57.9
Qwen3-VL-8B-Instruct 2025.12		53.9
Qwen3-VL-4B-Instruct 2025.12		51.6
Gemini-2.0-Pro 2025.12		48.1
InternVL3.5-8B-Instruct 2025.12		46.4
InternVL3.5-8B-MPO 2025.12		46.4
Keye-VL-1.5-8B 2025.12		46
Claude-Sonnet-3.7 2025.12		41.9
Qwen2.5-VL-7B + CARE 2025.12		32.6
Qwen2.5-VL-7B + GSPO 2025.12		31.6
M2-Reasoning 2025.12		31.5
Qwen2.5-VL-7B + DAPO 2025.12		29.4
DeepEyes 2025.12		26.6
Qwen2.5-VL-7B + GRPO 2025.12		25.7
LLaVA-OneVision-1.5 8B 2025.12		25.6
Qwen2.5-VL-3B + CARE 2025.12		25
Qwen2.5-VL-7B 2025.12		22.4
Qwen2.5-VL-3B 2025.12		21.2