Vision-Centric Reasoning on V* Bench (Overall) benchmark leaderboard · SOTA2 Research

Vision-Centric Reasoning on V* Bench (Overall)

96.5Attribute ScoreQwen3-VL-4B + SD-RPN

Evaluation Results

Method	Links
Qwen3-VL-4B + SD-RPN2026.04		96.5	82.9	91.1
ZwZ-Qwen3-VL-4B + Q-Zoom2026.04		96.5	90.8	94.2
Qwen3-VL-4B + Q-Zoom2026.04		95.7	85.5	91.6
Qwen2.5-VL-7B + SD-RPN2026.04		94.8	77.6	88
ZwZ-Qwen2.5-VL-7B + Q-Zoom2026.04		94.8	86.8	91.6
Qwen2.5-VL-3B + SD-RPN2026.04		91.3	65.8	81.2
Qwen2.5-VL-7B + Q-Zoom2026.04		89.6	79	85.3
ZwZ-Qwen3-VL-4B2026.04		89.6	86.8	88.5
Qwen2.5-VL-3B + Q-Zoom2026.04		87	69.7	80.1
ZwZ-Qwen2.5-VL-7B2026.04		87	82.9	85.3
Qwen3-VL-4B2026.04		86.1	76.3	83.2
Qwen2.5-VL-7B + Thyme2026.04		83.5	80.3	82.2
Qwen2.5-VL-3B2026.04		81.7	60.5	73.3
Qwen2.5-VL-7B2026.04		80	75	78
LLaVA-1.5-7B + SD-RPN2026.04		70.4	71.1	70.7
LLaVA-1.5-7B + Q-Zoom2026.04		70.4	71.1	70.7
LLaVA-1.5-13B + Q-Zoom2026.04		61	64.5	61.8
LLaVA-1.5-13B + SD-RPN2026.04		60.9	65.8	62.8
LLaVA-1.5-7B + ViCrop2026.04		53.9	50	52.4
LLaVA-1.5-7B + S22026.04		53	59.1	55.5
LLaVA-1.5-7B2026.04		48.7	52.6	50.3
LLaVA-1.5-13B + ViCrop2026.04		47.8	57.9	51.8
LLaVA-1.5-13B2026.04		47	56.6	50.8
LLaVA-1.5-13B + S22026.04		43.5	59.2	49.7