Benchmark RadarAI BENCHMARK PROFILE
Video-IFBench
Evaluates instruction following of multimodal LLMs in video understanding with 1.5K samples across constraint categories.
- Released
- 2026-08-26
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Addresses the underexplored area of instruction adherence in video understanding, where user-specified constraints are crucial.
Motivation
Multimodal Large Language Models (MLLMs) have shown strong performance in video understanding.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.