VMTP or Visual Media Transcription Protocol is a video processing protocol for LLMs. Using this, you can let your LLMs & AI agents understand video input.