Voxtral Small 24B

Voxtral Small 24B is Mistral's lightweight entry in the Voxtral family, designed to handle both text and audio processing tasks. As a 24 billion parameter model, it sits in the lightweight tier of Mistral's model lineup, offering multimodal capabilities at a smaller scale than flagship alternatives. The model features a 32,000 token context window and supports both text and audio inputs, enabling applications that require understanding of spoken content alongside written text. This dual-modality approach allows for transcription, audio analysis, and cross-modal reasoning tasks without requiring separate specialized models. Voxtral Small 24B targets use cases where audio processing is needed but computational efficiency and cost control are priorities. It provides an alternative to larger multimodal models when the specific combination of text and audio understanding is required at scale.