GLM 4.6V Flash

GLM 4.6V Flash is the smaller sibling in Z.ai's GLM 4.6V vision-language family, designed for production multimodal use rather than research demos. It is released under an open-source MIT license and is distributed as zai-org/GLM-4.6V-Flash on Hugging Face, giving developers direct access to the weights. The model targets workflows that blend visual understanding with conversational reasoning, positioning it as a deployable alternative to larger closed vision-language systems.

The 9B-parameter Flash variant is engineered to run on a single high-end GPU while retaining the multimodal breadth of its 106B counterpart. It handles image, video, and text inputs together in a shared 128,000-token context window and supports native tool calling for agent-style pipelines. This combination of compact size, open weights, and long mixed-modality context makes it well suited for document understanding, video analysis, and visual question answering in environments where self-hosting and customization matter.