Multimodal Knowledge and Math on MMMU (val)

69.1Accuracy

GPT-4o

Updated 20d ago

Evaluation Results

Method	Links
GPT-4o 2024.10		69.1
Gemini-1.5 Pro 2024.10		62.2
Ovis2.5 2026.01		60.6
GPT-4o mini 2024.10		59.4
Qwen3-VL 2026.01		57.1
GPT-4V 2024.10		56.4
Gemini-1.5 Flash 2024.10		56.1
ARIA 2024.10		54.9
Ostrakon-VL 2026.01		54.8
InternVL3.5 2026.01		54.4
GLM-4.6V-FlashX 2026.01		54.2
Pixtral-12B 2024.10		52.5
Qwen2.5-VL 2026.01		51.5
Llama3.2-11B 2024.10		50.7
LLaVA-OV-7B + HERMES 2026.05		48
Cambrian-4B + Gaze attention 2026.05		47.4
Qwen2.5-VL-3B 2026.05		46.6
Cambrian-4B 2026.05		46.4
Cambrian-4B + Gaze attention 2026.05		46.4
Cambrian-4B + Gaze attention 2026.05		46.3
LLaVA-OV-7B + HERMES 2026.05		46
Cambrian-4B + TokenCompre. 2026.05		45.9
Qwen2.5-VL-3B + InfiniPot-V 2026.05		45.8
Cambrian-4B + TokenCompre. + Gaze attention 2026.05		45.7
Cambrian-4B + TokenCompre. + Gaze attention 2026.05		45.3
LLaVA-OV-7B 2026.05		45.1
Qwen2.5-VL-3B + InfiniPot-V 2026.05		44.2
LLaVA-Next-LLaMA-3-8B 2026.05		39.6
Cambrian-8B-DINOv2-L 2026.05		36.3
LLaVA-Next-Qwen2-7B 2026.05		36.1
Cambrian-8B-CLIP-L 2026.05		36.1
SPHINX-7B 2026.05		31.6
BLIP-2-7B 2026.05		26.8