mistralai

/

Voxtral-Mini-3B-2507

like 529

Follow

Mistral AI_ 11.7k

I used vllm to run this model. After struggling, I made to work.
Couple things I didn't enjoy:

1. consolidated.safetensors is mandatory. (many other models work without it.)
2. It transcribes mono audio only. I have to convert stereo audio files into mono first. phi-4-multimodal-instruct model transcribes both stereo and mono.