6-bit MLX quantization of mistralai/Mistral-Small-3.2-24B-Instruct-2506, for Apple Silicon (~18 GB). Text-only build (vision tower not included) — run it with mlx-lm, not mlx-vlm. Ships a LlamaTokenizerFast tokenizer (vocab 131072) and chat template.

Usage

pip install -U mlx-lm

bash

mlx_lm.generate \
  --model TyKaoz/Mistral-Small-3.2-24B-Instruct-2506-6bit \
  --prompt "Explique la quantization en une phrase." \
  --max-tokens 200

Base	Tool	Precision	Size
`mistralai/Mistral-Small-3.2-24B-Instruct-2506`	`mlx-lm`	6-bit · group 64	~18 GB

By TyKaoz — privacy-first native macOS LLM chat client. Apache 2.0, inherited from the base model.