Base44 Superagents-AI agent that does it all
Mini toolsvideo SALMONN 2
By ByteDance
video-SALMONN 2 is a caption-enhanced audio-visual LLM that jointly processes video and audio to produce detailed captions and QA responses. Released in 3B, 7B and 72B sizes, it tops benchmarks like Video-MME, WorldSense and AVUT for audio-visual QA and performs strongly on visual-only tests such as MLVU and LVBench. The repo provides training and evaluation code plus checkpoints and an upgraded video-SALMONN 2+ line.
New
Text
Gen 4
Overview
video-SALMONN 2 is an audio-visual large language model from Tsinghua and ByteDance that uses video frames and sound to generate rich captions and answers, reaching state-of-the-art on many audio-visual QA and video understanding benchmarks.
About ByteDance
ByteDance is a multinational technology company known for its content platforms, including TikTok and Douyin.
View Company ProfileTools using video SALMONN 2
No tools found for this model yet.
Last updated: February 25, 2026
