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

Multimodal Visual Perception on MMVP

85.33Accuracy

GPT-5

Updated 19d ago

Evaluation Results

Method	Links
GPT-5 2025.12		85.33
GPT-4o 2025.12		84.66
InternVL2.5-78B 2025.12		83
Qwen3VL-32B 2025.12		82.66
Qwen2.5-VL-72B 2025.12		81.7
InternVL3.5-38B 2025.12		80.66
Qwen3-VL-4B-Thinking 2025.12		80
Qwen3-VL-4B-Thinking + DiG 2025.12		79
Qwen3-VL-8B-Thinking + DiG 2025.12		78.7
COOPER 2025.12		78.66
Qwen3VL-8B 2025.12		77.33
Qwen3-VL-8B-Thinking 2025.12		77.3
InternVL3.5-8B 2025.12		76.33
BAGEL-RE 2025.12		75.33
BAGEL 2025.12		74.33
Florence-VL 8B 2024.12		73.3
BAGEL-PE 2025.12		72.66
BAGEL 2025.11		69.3
Phi 3.5 Vision 2024.12		67.7
Florence-VL 3B 2024.12		64.7
AutoV 2025.06		64.3
AutoV 2025.06		64.2
Janus-Pro-7B 2025.12		62.33
API 2025.06		61.7
API 2025.06		61.5
Base 2025.06		61.3
Base 2025.06		60.7
Liquid-7B 2025.12		58.33
Qwen2.5-VL-7B 2025.12		54.7
Cambrian 8B 2024.12		51.3
Janus-Pro + AIA (Ours) 2025.11		48
Janus-Pro 2025.11		47.3
LLaVA next 8B 2024.12		38.7
CoVFT 2026.03		38.67
JARVIS 2025.12		38
BitFit 2026.03		38
FedAvg-TIEs 2026.03		37.3
Fed-CMP 2026.03		36.9
LLaVA 2025.12		36.7
CoVFT 2026.03		36.67
Emu3-Chat 2025.11		36.6
FedAvg-TA 2026.03		36.6
MOON 2026.03		36.4
VIRAL 2025.12		36
FedAvg-DARE 2026.03		36
Ours w/o Masking 2025.12		35.3
FedAvg 2026.03		35.2
Freeze 2026.03		34
Full fine-tuning 2026.03		34
FedAdam 2026.03		33.9
BitFit 2026.03		31.33
LLaVA 2025.12		31.3
JARVIS 2025.12		30.7
LLaVA-1.5 2026.03		30.67
Local Training 2026.03		30.5
VIRAL 2025.12		30
SVPT 2026.03		29.33
LLaVA-1.5-7B 2024.12		29.3
LoRA 2026.03		28.66
FedProx 2026.03		28.2
Ours w/o Masking 2025.12		28
Freeze 2026.03		28
LLaVA-1.5 2026.03		27.33
Full fine-tuning 2026.03		27.33
Dynamic-LLaVA-7B† 2024.12		26.3
VPT 2026.03		24.67
LLaVA-FastV 2024.12		24
VILA-U 2025.11		22
Mini-Gemini-HD 8B 2024.12		18.7
Emu3 + AIA (Ours) 2025.11		10.8
Emu3 2025.11		8.7
LLaVA-1.5-7B+H2O 2024.12		0