Mistral AI logo

Voxtral

Mistral AIOpen WeightsPending Human Review

Voxtral is Mistral’s open-weight audio-language family for understanding speech as well as transcribing it. The 3B Mini and 24B Small models answer questions, summarize recordings, and use tools directly from audio input, with multilingual support. Both are Apache-2.0 licensed and have hosted API access. The separately optimized Mini Transcribe service is a transcription-only product and is not a separate language model page here. Audio understanding generates text; this release is not a text-to-speech system.

2025-07-15
3B, 24B
Audio encoder with language-model backbone
Apache-2.0

Specifications

Parameters
3B, 24B
Architecture
Audio encoder with language-model backbone
License
Apache-2.0
Context Window
32,768 tokens
Type
multimodal
Modalities
textaudio

Benchmark Scores

Advanced Specifications

Model Family
Voxtral
API Access
Available
Chat Interface
Not Available
Multilingual Support
Yes
Variants
Voxtral Mini 3BVoxtral Small 24B

Capabilities & Limitations

Capabilities
audio question answeringsummarizationtranscriptionmultilingualfunction calling
Known Limitations
Audio durations must fit supported limitsProduces text rather than spoken output
Notable Use Cases
meeting analysisvoice assistantsaudio search
Function Calling Support
Yes

Related Models