InclusionAI's Ling-3.0-flash-VL-fp4 repository provides a downloadable model that accepts images and video alongside text, so it can be asked questions about visual material in the same conversation. The model card shows an image request example for a self-hosted server and the message format for video input. Model card.
The release gives developers a documented route to serving the model themselves. The card's recommended deployment recipe targets two large GPUs, so the downloadable weights should not be confused with a model demonstrated to run on an ordinary laptop.
Terms in this piece · Glossary
mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
