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

General Perception and Reasoning on MME-RealWorld Lite

54.3Overall Accuracy

Deepeyes

Updated 21d ago

Evaluation Results

Method	Links
Deepeyes 2026.03		54.3	-	-	-	-	-	-	-	-	-	50.5	56.6
DeepEyes 2026.05		54.3	-	-	-	-	-	-	-	-	-	50.5	56.6
V-Reflection 2026.03		53.9	-	-	-	-	-	-	-	-	-	45	58.5
DeepEyes 2025.05		53.2	90	52.7	89	43.3	33.4	76	69	44	35	-	-
GPT-4o 2026.03		52	-	-	-	-	-	-	-	-	-	48.3	54.4
GPT-4o 2026.05		52	-	-	-	-	-	-	-	-	-	48.3	54.4
UniVLR 2026.05		50.7	-	-	-	-	-	-	-	-	-	44.7	54.5
Qwen2.5-VL-7B 2026.03		50.1	-	-	-	-	-	-	-	-	-	42.5	54.1
Pixel-Reasoner 2025.05		49.7	89.6	52	86	38.9	30.9	71	72	46	32.5	-	-
PixelReasoner 2026.05		49.7	-	-	-	-	-	-	-	-	-	44.5	53.1
InternVL3 2026.03		48.6	-	-	-	-	-	-	-	-	-	44.8	51
CoVT 2026.05		48.2	-	-	-	-	-	-	-	-	-	42.9	51.6
Monet 2026.05		46.9	-	-	-	-	-	-	-	-	-	40.3	51.2
Qwen2.5-VL-7B 2026.05		46.2	-	-	-	-	-	-	-	-	-	43.1	48.2
Qwen2.5-VL-7B 2026.03		45.8	-	-	-	-	-	-	-	-	-	39.7	49.6
Qwen2.5-VL 2025.05		45.6	87.2	40.7	83	29.5	40.7	74	60	27.3	29.5	-	-
SkiLa 2026.05		45.6	-	-	-	-	-	-	-	-	-	36.3	51.6
Qwen2.5-VL-7B + vanilla SFT 2026.05		45.5	-	-	-	-	-	-	-	-	-	39.9	49.1
LLaVA-OneVision 2025.05		43.7	80	40	56	31.7	39.4	65	33	38	32	-	-
Qwen2.5-VL 2025.05		42.3	87.6	32.7	83	27.3	30	72	62	28.7	23	-	-
LVR 2026.05		39.1	-	-	-	-	-	-	-	-	-	37.5	40.1