List your company
ResearchFormerly WizWand

Multimodal Reasoning on MATH-Vision (full)

62.7AccuracyQwen3-VL-8B-Thinking
14.44426.97239.552.028Jul 9, 2025Aug 5, 2025Sep 2, 2025Sep 30, 2025Oct 27, 2025Nov 24, 2025Dec 22, 2025
Updated 5mo ago

Evaluation results

MethodLinks
Qwen3-VL-8B-ThinkingModel Type=Reasoning2025.1262.7
Qwen3-VL-8B-Instruct + CAREBackbone=Qwen3-VL-8B-Instruct, Method=CARE2025.1261.7
MiMo-VL-7B-RLModel Type=Reasoning2025.1260.4
Qwen3-VL-4B-ThinkingModel Type=Reasoning2025.1260
GPT-5-NanoModel Type=Proprietary2025.1259.7
MiMo-VL-7B-SFTModel Type=Instruct2025.1257.9
Qwen3-VL-8B-InstructModel Type=Instruct2025.1253.9
Qwen3-VL-4B-InstructModel Type=Instruct2025.1251.6
Gemini-2.0-ProModel Type=Proprietary2025.1248.1
InternVL3.5-8B-InstructModel Type=Instruct2025.1246.4
InternVL3.5-8B-MPOModel Type=Reasoning2025.1246.4
Keye-VL-1.5-8BModel Type=Reasoning2025.1246
Claude-Sonnet-3.7Model Type=Proprietary2025.1241.9
Qwen2.5-VL-7B + CAREBackbone=Qwen2.5-VL-7B, Method=CARE2025.1232.6
Qwen2.5-VL-7B + GSPOBackbone=Qwen2.5-VL-7B, RL Method=GSPO2025.1231.6
M2-ReasoningModel Type=Reasoning2025.1231.5
GPT-4o2025.0730.3
Qwen2.5-VL-7B + DAPOBackbone=Qwen2.5-VL-7B, RL Method=DAPO2025.1229.4
D2IparReasoning Strategy=D2I, Variant=par, Backbone=Qwen2.5-VL-7B2025.0727.2
DeepEyesModel Type=Reasoning2025.1226.6
D2IlocReasoning Strategy=D2I, Variant=loc, Backbone=Qwen2.5-VL-7B2025.0726.3
Qwen2.5-VL-7B + GRPOBackbone=Qwen2.5-VL-7B, RL Method=GRPO2025.1225.7
LLaVA-OneVision-1.5 8BModel Type=Instruct2025.1225.6
OpenVLThinker-7B2025.0725.3
Qwen2.5-VL-3B + CAREBackbone=Qwen2.5-VL-3B, Method=CARE2025.1225
GPT-4V2025.0722.7
Qwen2.5-VL-7B w/ GRPOReasoning Mode=Deliberate, Backbone=Qwen2.5-VL-7B2025.0722.5
Qwen2.5-VL-7BModel Type=Instruct2025.1222.4
Qwen2.5-VL-7B*Backbone=Qwen2.5-VL-7B2025.0721.3
Qwen2.5-VL-3BModel Type=Instruct2025.1221.2
D2DjusReasoning Strategy=D2D, Variant=jus, Backbone=Qwen2.5-VL-7B2025.0720.1
D2DlocReasoning Strategy=D2D, Variant=loc, Backbone=Qwen2.5-VL-7B2025.0720
InternVL2.5-8B2025.0719.7
D2IjusReasoning Strategy=D2I, Variant=jus, Backbone=Qwen2.5-VL-7B2025.0719.7
Qwen2.5-VL-7B w/ GRPO†Reasoning Mode=Intuitive, Backbone=Qwen2.5-VL-7B2025.0718.8
InternVL2-8B2025.0718.4
D2DparReasoning Strategy=D2D, Variant=par, Backbone=Qwen2.5-VL-7B2025.0717.6
Qwen2-VL-7B2025.0716.3