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

Multi-discipline Multimodal Understanding on MMMU

84.2Accuracy

GPT-5

Updated 6d ago

Evaluation Results

Method
GPT-5 2025.12	84.2	-	-	-	-	-
GPT-5-Mini 2025.12	79	-	-	-	-	-
InternVL3.5-38B 2025.12	76.9	-	-	-	-	-
InternVL3.5-38B 2025.06	76.9	82.8	-	-	-	-
GenRecal (InternVL3.5-8B) 2025.06	76.5	83.4	-	-	-	-
GenRecal (InternVL3.5-8B) 2025.06	76.3	83.4	-	-	-	-
Qwen3-VL-32B 2025.12	76	-	-	-	-	-
Qwen3-VL-32B 2025.06	76	82.5	-	-	-	-
GLM-4.5V 2025.12	75.4	-	-	-	-	-
Gemini-2.5-Pro 2025.12	74.7	-	-	-	-	-
Claude-4-Sonnet 2025.12	74.4	-	-	-	-	-
InternVL3-8B-Masters 2025.12	74	-	-	-	-	-
GPT-4.1 2025.12	74	-	-	-	-	-
Masters 2025.12	74	-	-	-	-	-
InternVL3.5-8B 2025.12	73.4	-	-	-	-	-
InternVL3.5-8B 2025.06	73.4	80.4	-	-	-	-
Qwen3-VL-8B-Masters 2025.12	72.9	-	-	-	-	-
Masters 2025.12	72.9	-	-	-	-	-
InternVL3.5-8B-Masters 2025.12	72.7	-	-	-	-	-
Masters 2025.12	72.7	-	-	-	-	-
GenRecal (Qwen3-VL-8B) 2025.06	72.7	80.1	-	-	-	-
GenRecal (Qwen3-VL-8B) 2025.06	72.5	80	-	-	-	-
InternVL3-78B 2025.12	72.2	-	-	-	-	-
Keye-VL-8B 2025.12	71.4	-	-	-	-	-
Keye-VL-1.5-8B 2025.12	71.4	-	-	-	-	-
Qwen2.5-VL-7B-Masters 2025.12	71.3	-	-	-	-	-
Masters 2025.12	71.3	-	-	-	-	-
Claude-3.7-Sonnet 2025.12	71	-	-	-	-	-
Qwen3-VL-4B-Masters 2025.12	70.3	-	-	-	-	-
Gemini-2.0-Flash 2025.12	69.9	-	-	-	-	-
Qwen3-VL-8B 2025.12	69.6	-	-	-	-	-
Qwen3-VL-8B 2025.06	69.6	77	-	-	-	-
VLsI-7B 2024.12	69.3	-	-	-	-	-
GPT-4o 2024.09	69.2	-	-	-	-	-
GPT-4o 2025.12	69.1	-	-	-	-	-
Claude-3.5-Sonnet 2025.12	68.3	-	-	-	-	-
Qwen2.5-VL-72B 2025.12	68.2	-	-	-	-	-
GLM-4.1V-9B 2025.12	68	-	-	-	-	-
Qwen3-VL-4B 2025.12	67.4	-	-	-	-	-
MiMo-VL-8B 2025.12	66.7	-	-	-	-	-
InternVL3.5-4B 2025.12	66.6	-	-	-	-	-
InternVL3.5-2B-Masters 2025.12	64.6	-	-	-	-	-
InternVL3.5-4B-Masters 2025.12	63.2	-	-	-	-	-
GPT-4o 2024.06	62.8	-	-	-	-	-
SketchThinker-R1-7B 2026.01	62.8	-	-	64.3	97.7	-
InternVL3-8B 2025.12	62.7	-	-	-	-	-
Gemini-1.5-Pro 2025.12	62.2	-	-	-	-	-
Qwen2.5-VL-3B-Masters 2025.12	61.2	-	-	-	-	-
Vanilla-R1 2026.01	61	-	-	182.2	33.5	-
VeriThinker 2026.01	60.1	-	-	105.8	56.8	-
EMOVA 2024.09	59.7	-	-	-	-	-
NVLM-72B 2025.12	59.7	-	-	-	-	-
InternVL3-2B-Masters 2025.12	59.6	-	-	-	-	-
L1 2026.01	59.5	-	-	136.8	43.5	-
C3oT 2026.01	59.3	-	-	127.1	46.7	-
ThinkPrune 2026.01	59.2	-	-	104.9	56.4	-
InternVL3.5-2B 2025.12	59	-	-	-	-	-
Chain-of-Draft 2026.01	58.9	-	-	86.3	68.3	-
Constrained CoT 2026.01	58.6	-	-	78.2	74.9	-
Gemini Pro 1.5 2024.09	58.5	-	-	-	-	-
Ovis2-8B 2025.12	57.4	-	-	-	-	-
VisionZero-CLEVR 2026.02	56.98	-	-	-	-	-
GPT-4V 2024.01	56.8	-	-	-	-	-
GPT-4V 2024.03	56.8	-	-	-	-	-
GPT-4V 2024.07	56.8	-	-	-	-	-
LLaVA-OneVision 2024.09	56.8	-	-	-	-	-
GPT-4V 2024.09	56.8	-	-	-	-	-
LLaVA-OneVision-72B 2025.12	56.8	-	-	-	-	-
Active-Zero 2026.02	56.42	-	-	-	-	-
Oryx-1.5 2024.09	56.1	-	-	-	-	-
Phi-4-Multimodal-5.6B 2025.12	56	-	-	-	-	-
Qwen3-VL-2B-Masters 2025.12	55.8	-	-	-	-	-
LLaVA-OneVision-1.5-8B 2025.12	55.4	-	-	-	-	-
EvoLMM 2026.02	55.31	-	-	-	-	-
Qwen2.5-VL-7B 2025.12	55	-	-	-	-	-
Qwen2-VL-7B 2024.12	54.1	-	-	-	-	-
Molmo-72B 2025.12	54.1	-	-	-	-	-
GPT-4v 2024.06	53.8	-	-	-	-	-
Base Model 2026.02	53.52	-	-	-	-	-
360VL 2024.07	53.4	-	-	-	-	-
Qwen3-VL-2B 2025.12	53.4	-	-	-	-	-
Prism Captioner-7B 2024.06	53.3	-	-	-	-	-
VisionZero-RealWorld 2026.02	53.3	-	-	-	-	-
VisionZero-Chart 2026.02	53.07	-	-	-	-	-
LLaVA-OneVision-1.5-4B 2025.12	52.7	-	-	-	-	-
VisPlay 2026.02	52.51	-	-	-	-	-
VITA 2024.09	52.1	-	-	-	-	-
VILA-1.5 2024.09	51.9	-	-	-	-	-
Qwen-VL-MAX 2024.03	51.4	-	-	-	-	-
Qwen2.5-VL-7B + SaFeR-VLM 2026.03	51.33	-	-	-	-	-
Phantom-7B 2024.12	51.2	-	-	-	-	-
Qwen2.5-VL-3B 2025.12	51.2	-	-	-	-	-
Qwen2.5-VL-7B 2026.03	51.11	-	-	-	-	-
SpaRE-7B 2025.04	51	-	-	-	-	-
Qwen2VL-7B 2025.04	51	-	-	-	-	-
MiniCPM-o2.6-8B 2025.12	50.9	-	-	-	-	-
Qwen2.5-VL-7B + SPA-VL 2026.03	50.33	-	-	-	-	-
Oryx 2024.09	50.3	-	-	-	-	-
TroL-7B 2024.12	49.9	-	-	-	-	-
Qwen2.5-VL-7B + SFT+DPO+GRPO 2026.03	49.89	-	-	-	-	-

Showing 100 of 499 rows