V*Bench

The 'V* Bench' dataset/benchmark is used to evaluate the performance and efficiency of agentic multimodal large language models (MLLMs) in tasks involving visual perception and reasoning.

Papers using V*Bench (3)

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning2026Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models2025Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models2025