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

Home/Benchmarks

Visual Reasoning on V* Bench (test)

91.3Accuracy

DeepEyes

Updated 1mo ago

Evaluation Results

Method	Links
DeepEyes 2025.07		91.3
MGPO 2025.07		76.4

SOTA Paper

DeepEyes

DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning

Dataset

V*Bench

Follow for update

@wizwand_team Discord

Related Benchmarks

High-Resolution Visual Reasoning on HR-Bench 8K High-Resolution Visual Reasoning on HR-Bench 4K Fine-grained Visual Perception on V* Bench (OOD)Multimodal Perception and Reasoning on MME-Realworld ID Multi-modal Understanding on MME-Realworld

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs