mlx-community

/

GLM-4.5V-8bit

like 2

Follow

MLX Community 7.85k

It seems that it is still in progress.
See https://github.com/Blaizzy/mlx-vlm/commits/pc/glm-4-5v/

This is an LM Studio issue rather than an MLX_LM/MLX-VLM issue. LM Studio hasn’t yet added support for the glm4v_moe model type. Vision models generally take longer to get support than language-only models.

Also, as hehua2008 mentioned, the MLX-VLM commit for glm4v_moe is still in progress. I’d consider available quants as a hack rather than something stable. A little surprised it’s linked to MLX-community already.