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

Mathematical Multimodal Reasoning on MathVerse

57.6Accuracy

Qwen2.5-VL-72B

Updated 4d ago

Evaluation Results

Method	Links
Qwen2.5-VL-72B 2025.12		57.6
VisonR1 7B 2026.01		52.4
Ours 2026.01		49.4
Ours [with DPS and annealing] 2026.01		48.5
VLAA-Thinker 7B 2026.01		47.8
Ours [with DPS] 2026.01		47.6
DeepEyes-7B 2025.12		47.3
CodeDance-7B 2025.12		46.8
R1-Onevision 7B 2026.01		46.4
Qwen2.5-VL-7B 2025.12		45.1
ARM-Thinker-7B 2025.12		41.6
Qwen2.5-VL-7B 2025.12		41.1
MixedR1 7B 2026.01		40.8
Qwen2.5VL 7B 2026.01		40.4
R1-Onevision-7B 2025.12		40
R1-VL-7B 2025.12		40
InternVL3-8B 2025.12		39.8
InternVL2.5 8B 2026.01		39.5
InternVL2.5-8B 2025.12		39.5
GPT-4V 2026.01		39.4
GPT-4o 2025.12		35.3
Gemma-3-27B 2025.12		34
LLaVA-OneVision 7B 2026.01		26.2
Llava-OV-7B 2025.12		26.2
Fused 2026.02		19.29
MaD-Mix 2026.02		18.91
Avg 2026.02		18.32
Uniform 2026.02		18.27
Human 2026.02		17.26