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

Multimodal Perception on HR-4K

87.88Accuracy

Gemini-3-Flash

Updated 26d ago

Evaluation Results

Method	Links
Gemini-3-Flash 2026.02		87.88
Qwen3-VL-235B-A22B 2026.02		84.5
ZwZ-8B 2026.02		84.38
Gemini-2.5-Pro 2026.06		83.9
TACO (Ours-7B) 2026.06		83.8
Kimi-K2.5 (1T) 2026.02		81.87
ZwZ-4B 2026.02		81.75
GLM-4.5V (108B) 2026.02		81.63
Qwen3-VL-8B 2026.06		78.9
Qwen3-VL-8B 2026.02		78.88
Qwen3-VL-4B 2026.02		78.25
PyVision-RL-7B 2026.06		78.1
DeepEyes-v2-7B 2026.06		77.9
Mini-o3-7B 2026.06		77.5
Thyme-7B 2026.06		77
CodeV-RL-7B 2026.06		76.1
ZwZ-7B 2026.02		75.38
GPT-5 2026.06		75.3
DeepEyes-7B 2026.06		75.1
MiMo-VL-7B-RL 2026.02		74.38
Pixel-Reasoner-7B 2026.06		74
MathCoder-VL-8B 2026.06		73.8
Qwen2.5-VL-32B 2026.06		73.4
Qwen2.5-VL-7B 2026.02		71.62
InternVL3-8B 2026.06		70
MiniCPM-V-4.5 (9B) 2026.02		69.88
Qwen2.5-VL-7B 2026.06		68.8
GPT-5.1 2026.02		67
GPT-4o 2026.06		65
LLaVA-OneVision-7B 2026.06		63