Mistral Launches Voxtral, Its First Open Audio AI Model for Businesses
By James Dargan · · 1 min read
French AI startup Mistral has entered the speech AI arena with the launch of Voxtral, a new family of open-weight audio models designed to bring advanced speech intelligence to businesses at a fraction of the cost of closed systems. The release marks Mistral’s latest push to challenge proprietary AI incumbents by offering developers greater affordability and control.
Built on the company’s Mistral Small 3.1 LLM, Voxtral supports transcription, multilingual understanding, summarization, and voice command execution. Two versions are available: the 24B-parameter Voxtral Small for large-scale use, and the 3B-parameter Voxtral Mini for local and edge applications. A stripped-down Voxtral Mini Transcribe version targets high-speed, low-cost transcription tasks.
The open models are available via Hugging Face and Mistral’s Le Chat, with API access starting at $0.001 per minute. The launch follows the recent debut of Magistral, Mistral’s reasoning-focused AI model family, and comes amid reports that the company is in talks to raise $1 billion in new funding.
