MOSS-VL-Instruct-0708

by openmoss-teamco.huggingface.openmoss-team/moss-vl-instruct-0708-demo

MOSS_VL_Instruct_0708_Demo_analyze_image Analyze an image with a text prompt.
MOSS_VL_Instruct_0708_Demo_analyze_text Answer a text-only prompt (no media attached).
MOSS_VL_Instruct_0708_Demo_analyze_video Analyze a video with a text prompt.
MOSS_VL_Instruct_0708_Demo_chat_generate One multi-turn chat step: prior session messages + this turn -> reply. Built on the model's offline_batch_generate session API. `session` is a plain list of message dicts (media referenced by file path), so it round-trips through gr.State across stateless ZeroGPU calls; every turn re-encodes the full context including earlier media. Returns: {"text": assistant reply, "session": updated message history to pass next turn}.

As the server's own listing gives them. We never call a tool.

Tools