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

Multimodal Understanding and Reasoning on MMBench Chinese (dev)

72.8Accuracy

Deepseek-VL-7B

Updated 1mo ago

Evaluation Results

Method	Links
Deepseek-VL-7B 2026.01		72.8
LLaVA-FA-7B 2026.01		69.5
LLaVA-FA-3B 2026.01		68
MiniCPM-V-2 2026.01		67.2
LLaVA-NeXT 2026.01		64.4
MiniCPM-V 2026.01		62.7
VILA-7B 2026.01		61.7
LLaVA-FA-2B 2026.01		61.7
Imp-2B 2026.01		61.2
DeepSeek-VL-1.3B 2026.01		61
Bunny-2B 2026.01		58.5
LLaVA-1.5-7B 2026.01		58.3
MoE-LLaVA-2B 2026.01		57.3
Qwen-VL-Chat 2026.01		56.7
CogVLM 2026.01		53.8
VILA-3B 2026.01		52.7
Mini-Gemini-2B 2026.01		51.3
LLaVA-FA-1B 2026.01		49.4
Imp-3B 2026.01		46.7
MoE-LLaVA-3B 2026.01		41.8
SPHINX-Tiny 2026.01		37.8
Bunny-3B 2026.01		37.2