---
license: apache-2.0
---

https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503

Q6_K_XL: Q6_K weights, untouched outputs, untouched embed

Fits 24K CTX on a 24GiB GPU

llama.cpp does not support exporting the vision components yet