Mistral Voxtral 3

★★★★☆4.2 / 5 editorial Paid Pay-as-you-go: voice agent API $0.04/min with Mistral Small 4 reasoning ($0.08/min with another reasoning model); transcription from $0.0035/min; TTS $13 per 1M characters — check Mistral's pricing page for current rates.

Full voice-agent stack: transcription, expressive speech, and tool-using agents in 50+ languages.

Visit Site ♡ 2,597 saves

#voice-agents#transcription#text-to-speech#sdk#multilingual

What is Mistral Voxtral 3?

Mistral Voxtral 3, released October 20, 2026, is a complete voice-agent stack: Voxtral Small and Mini Transcribe 3 for batch transcription, Voxtral Mini Transcribe Realtime 3 for live calls, Voxtral Mini TTS 3 (Speech 3) for expressive speech, and a LiveKit-based voice agent SDK that streams listening, reasoning, tool calling, and speaking as a single pipeline with sub-second turns. Agents can look up a bill, book an appointment, or check a system mid-call, over SIP telephony or WebRTC in the browser.

Transcription covers 50+ languages with speaker diarization, context biasing for domain vocabulary, and end-of-turn detection, while speech generation spans 35+ languages and 50+ dialects with emotion, pacing, and emphasis controls plus Pro Voice cloning from 30 minutes of audio. It is available now through Mistral's API on EU, US, and global endpoints or via self-deployment, and because each stage is a separate model, teams can swap the language model, voice, or language without rebuilding the agent.

Key Features

Full voice-agent SDK on LiveKit: listen, reason, call tools, and speak in one streaming pipeline
Transcription in 50+ languages with diarization, context biasing, and end-of-turn detection
Expressive TTS in 35+ languages and 50+ dialects with emotion, pacing, and emphasis controls
Pro Voice cloning that adapts a speaker's character and style across languages from 30 minutes of audio
Swappable stages and EU-hosted API or self-deployment for data-sovereignty control

Pros

One vendor for the whole voice pipeline instead of three separate contracts
Strong multilingual transcription and diarization at a low per-minute price
Sub-second turns with real interruption handling, not push-to-talk

Cons

Models are in public preview, not general availability, so behavior and pricing can still change
Realtime transcription accuracy trails Gemini 3.5 Realtime at low-latency settings
No open weights were announced, and the full agent bill adds reasoning-model and telephony costs on top

Pricing

Model: Paid
Pay-as-you-go: voice agent API $0.04/min with Mistral Small 4 reasoning ($0.08/min with another reasoning model); transcription from $0.0035/min; TTS $13 per 1M characters — check Mistral's pricing page for current rates.

Adding Up Your AI Bill?

If you pay for Mistral Voxtral 3 alongside other AI subscriptions, the monthly total adds up quietly. Tick every tool you pay for in our free AI Tool Cost Calculator to see your real monthly and yearly bill — and spot overlaps you can cut. Still choosing? Start with our guide on how to choose the right AI tool.