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

Visual Question Answering on V*Bench

98.95Accuracy

Human

Updated 10d ago

Evaluation Results

Method
Human 2023.12	98.95	98.26	100
ConFoThinking (Qwen3-VL-8B) 2026.02	92.1	94.8	88.1
ConFoThinking (Qwen3-VL-4B) 2026.02	91.1	93.9	86.8
ZoomEye 2026.02	90.6	93.9	85.5
FOCUS 2025.06	90.58	-	-
ConFoThinking (Qwen2.5-VL-7B) 2026.02	86.9	90.4	81.5
UG-Search 2025.10	86.9	-	-
Qwen3-VL-8B (with tools) 2026.02	86.4	88.7	85.5
ZoomEye 2025.10	85	-	-
Qwen3-VL-8B 2026.02	84.3	86.1	81.6
Pixel-Reasoner 2026.02	84.3	85.2	82.9
Qwen3-VL-4B (with tools) 2026.02	83.2	85.2	80.3
UG-Search 2025.10	81.7	-	-
DyFo 2026.02	81.2	80	82.9
Visual Sketchpad 2026.02	80.1	81.7	77.6
InternVL3.5-8B 2026.02	79.1	82.6	73.7
Qwen-2.5-VL 2025.06	79.06	-	-
Qwen3-VL-4B 2026.02	78	80	75
LLaVA-OneVision-1.5-8B 2026.02	78	79.1	76.3
Qwen2.5-VL-7B 2026.02	76.4	78.2	73.6
SEAL 2023.12	75.39	74.78	76.31
SEAL 2026.02	74.9	78.3	68.7
LLaVA-OV-7B 2025.10	74.4	-	-
ViCrop 2025.10	69.6	-	-
Thyme 2025.10	68.1	-	-
TextCoT 2025.10	67	-	-
Qwen2.5-VL-7B 2025.10	64.9	-	-
ViCrop 2026.02	62.3	65.2	57.9
ICoT 2026.02	60.7	63.5	56.6
GPT-4V 2023.12	54.97	51.3	60.52
LLaVA-1.5 2023.12	48.68	43.47	56.57
Gemini Pro 2023.12	48.16	40.86	59.21
Ours (PSP & VRG) 2026.04	46.6	-	-
Ours (PSP) 2026.04	46	-	-
Ours (PSP & VRG) 2026.04	46	-	-
Ours (PSP) 2026.04	45.5	-	-
Ours (PSP & VRG) 2026.04	45.5	-	-
Margin 2026.04	44.5	-	-
Low-conf 2026.04	44.5	-	-
Ours (PSP) 2026.04	44.5	-	-
DDCoT 2026.04	43.9	-	-
CCoT 2026.04	43.9	-	-
Margin 2026.04	43.4	-	-
Margin 2026.04	43.4	-	-
Low-conf 2026.04	43.4	-	-
CCoT 2026.04	42.9	-	-
Entropy 2026.04	42.9	-	-
Low-conf 2026.04	42.9	-	-
DDCoT 2026.04	42.4	-	-
CCoT 2026.04	42.4	-	-
Entropy 2026.04	42.4	-	-
Entropy 2026.04	41.8	-	-
MM-React 2023.12	41.36	34.78	51.31
Visprog 2023.12	41.36	31.3	56.57
DDCoT 2026.04	41.3	-	-
Otter 2023.12	38.74	26.95	56.57
Ours (PSP & VRG) 2026.04	38.7	-	-
MiniGPT-4 2023.12	38.22	30.43	50
Ours (PSP) 2026.04	38.2	-	-
Ours (PSP & VRG) 2026.04	38.2	-	-
Ours (PSP) 2026.04	37.7	-	-
Ours (PSP & VRG) 2026.04	37.7	-	-
BLIP2 2023.12	37.69	26.95	53.94
VisualChatGPT 2023.12	37.69	30.43	48.68
Bard 2023.12	37.17	31.3	46.05
Ours (PSP) 2026.04	37.1	-	-
CCoT 2026.04	36.6	-	-
DDCoT 2026.04	36.1	-	-
CCoT 2026.04	36.1	-	-
CCoT 2026.04	36.1	-	-
Entropy 2026.04	36.1	-	-
Random Guess 2023.12	35.99	26.73	50
Entropy 2026.04	35.6	-	-
Margin 2026.04	35.6	-	-
Low-conf 2026.04	35.6	-	-
LLaVA 2023.12	35.59	23.47	53.94
DDCoT 2026.04	35	-	-
Entropy 2026.04	35	-	-
Low-conf 2026.04	35	-	-
DDCoT 2026.04	34.5	-	-
Margin 2026.04	34.5	-	-
Low-conf 2026.04	34.5	-	-
InstructBLIP 2023.12	34.02	25.21	47.36
Margin 2026.04	34	-	-