These Mistral 24B models just got released (hopefully not only based on reasoning), just trying to see if they're better than 22b ones and even L3.3 70B's,

if you can, please eval these ones in UGI:

https://huggingface.co/mistralai/Mistral-Small-24B-Instruct-2501
https://huggingface.co/mistralai/Mistral-Small-24B-Base-2501

https://huggingface.co/Mathieu-Thomas-JOSSET/MISTRAL-PATH

^ some custom-finetuned math/pathfinding based 24B I saw?