checklist.day

Voxtral Mini 4B Realtime 2602

JSON →
mistral audio
audiotext

A compact 4-billion-parameter real-time speech model from Mistral AI for low-latency voice applications.

Specs

context window 8K tokens
max output 8K tokens

Capabilities

streamingtool-usefunction-calling

Dates

releasedFeb 2025

Resources

homepagehuggingface.co/mistralai/Voxtral-Mini-4B-Realtime-2602 ↗

API

full doc /v1/models/mistralai/Voxtral-Mini-4B-Realtime-2602