Benchmark RadarAI BENCHMARK PROFILE
MME-RealWorld
A comprehensive evaluation benchmark for Multimodal Large Language Models featuring over 13,366 high-resolution images and 29,429 question-answer pairs across 43 subtasks and 5 real-world scenarios. The largest manually annotated multimodal benchmark to date, designed to test MLLMs on challenging high-resolution real-world scenarios.
- Released
- Unknown
- Readiness
- Paper only
- Primary field
- General AI
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.