# News: Mistral launches Voxtral 3 voice agent SDK from $0.04 per minute

> Mistral launched Voxtral 3 on 20 October 2026: three speech-to-text models, a text-to-speech model, and a voice agent SDK built on LiveKit that runs listening, reasoning, tool calling and speaking as one streaming pipeline. The voice agent API is priced at $0.04 per minute with Mistral Small 4 as the reasoning model and $0.08 per minute with a larger reasoning model. Transcription starts at $0.0035 per minute for Voxtral Mini Transcribe 3 in batch, $0.004 for Small Transcribe 3 and $0.008 for Mini Transcribe Realtime 3, and text-to-speech is $13 per million characters. Mistral claims transcription across more than 50 languages with diarisation and end-of-turn detection, realtime latency under 500 milliseconds, speech output in more than 35 languages and 50+ dialects, and availability through its EU-hosted API or self-deployment on customer infrastructure.

Human page: https://testedactually.com/news/mistral-voxtral-3-voice-agents/
Mirror: https://testedactually.com/news/mistral-voxtral-3-voice-agents.md
Published: 2026-10-23

## What changed

Mistral launched Voxtral 3 on 20 October 2026: three speech-to-text models, a text-to-speech model, and a voice agent SDK built on LiveKit that runs listening, reasoning, tool calling and speaking as one streaming pipeline. The voice agent API is priced at $0.04 per minute with Mistral Small 4 as the reasoning model and $0.08 per minute with a larger reasoning model. Transcription starts at $0.0035 per minute for Voxtral Mini Transcribe 3 in batch, $0.004 for Small Transcribe 3 and $0.008 for Mini Transcribe Realtime 3, and text-to-speech is $13 per million characters. Mistral claims transcription across more than 50 languages with diarisation and end-of-turn detection, realtime latency under 500 milliseconds, speech output in more than 35 languages and 50+ dialects, and availability through its EU-hosted API or self-deployment on customer infrastructure.

## What it means for buyers

A complete voice agent at $0.04 per minute is direct price pressure on the per-minute build platforms in our ai-voice-agents cluster, Vapi and Bland, and the EU-hosted or self-deployed option is a differentiator for buyers with data-residency requirements. The pipeline design keeps a text transcript of every turn, which matters for contact-centre audit. Two caveats for buyers: the model weights were not open at launch, so self-deployment here means Mistral-managed infrastructure rather than an open download, and every price above is vendor-stated launch pricing, not a verified invoice.

## Sources

- [1] Voxtral 3: voice agent SDK and speech models (Mistral newsroom) - https://mistral.ai/news/voxtral-3 (company announcement, model lineup, SDK architecture and launch pricing, 20 October 2026, checked 2026-10-23)
- [2] Voxtral 3: EU transcription for cents (Kiberblick) - https://kiberblick.de/aktuell/2026/2026-10-21-mistral-voxtral-3-transkription-eu (independent write-up with per-model pricing table and benchmark comparison, 21 October 2026, checked 2026-10-23)

TestedActually tracks pricing and programme changes as dated, sourced events: https://testedactually.com/news/.