Nimbus-Labs/Nimbus-9B-v2.1-MLX-4bit

Rebuild release integrity metadata

fix chat_template: 9B defaults to reasoning (Qwen3.5 enable_thinking)

Update metadata with huggingface_hub

Replace with mlx-vlm 4-bit export (vision tower preserved)

Align MLX card with mlx-lm Hub conventions (library_name=mlx, no GGUF tagging)

Fix MLX model card and release metadata (was GGUF leftovers)

Upload folder using huggingface_hub

initial commit