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

Perception on HallusionBench

59.5Score

AVAR-Thinker

Updated 1mo ago

Evaluation Results

Method	Links
AVAR-Thinker 2026.03		59.5
Claude-3.7-Sonnet 2026.03		58.3
GPT-4o 2026.03		56.2
Vision-SR1 2026.03		54.3
Mulberry-7B 2026.03		54.1
OpenVLThinker 2026.03		53
ThinkLite-VL 2026.03		52.3
Vision-R1 2026.03		51.9
InternVL2.5-8B 2026.03		51.1
VLAA-Thinker-7B 2026.03		50.9
Qwen2.5-VL-7B 2026.03		50.7
MM-Eureka-7B 2026.03		50.7
LLaVA-OneVision-7B 2026.03		47.5
R1-OneVision 2026.03		46
Llama-3.2-11B-Vision-Instruct 2026.03		40.3