New model2026-10-20via Product HuntVoice & Text-to-Speech

Mistral Voxtral 3

Name: Mistral Voxtral 3
Author: Mistral AI

Voxtral 3 is Mistral's new family of speech models plus a voice agent SDK: Transcribe 3 speech-to-text in 50+ languages with diarization and context biasing, Mini TTS 3 expressive text-to-speech in 35+ languages, and a LiveKit-based SDK that chains listening, reasoning, tool calls and speech.

Visit Mistral Voxtral 3 ↗

Paid (usage-based)Proprietary API; self-deployment available (license terms not checked) (unverified)REST API / SDK; Modelby Mistral AI

Why it matters

For businesses building phone or web voice agents, such as contact centers and service desks, it bundles transcription, speech and tool-calling into one pipeline that can run on Mistral's EU-hosted API or on their own infrastructure.

Highlights

Voxtral Small Transcribe 3 handles 50+ languages with speaker diarization, keyphrase context biasing and end-of-turn detection; Mistral reports it beats Whisper, Scribe v2 and Gemini 3.5 Transcribe on its benchmarks.unverified
Voxtral Realtime Transcribe 3 targets sub-500 ms transcription in 35+ languages for live agents.
Voxtral Mini TTS 3 adjusts emotion, pace and emphasis from context, covers 35+ languages and 50+ dialects, and supports voice cloning from about 30 minutes of audio (Pro Voice).
The voice agent SDK is built on LiveKit and runs speech in, reasoning with a model like Mistral Small, tool calls and speech out as one streaming pipeline.
Published API prices: Mini Transcribe 3 batch $0.0035/min, Realtime $0.008/min, Small Transcribe 3 $0.0040/min, Mini TTS 3 $13 per 1M characters, voice agent API from $0.04/min.

Taken from the project's own README, docs or launch page. Claims about speed or accuracy are the maker's unless stated otherwise.

Try it if: You run a multilingual phone line or help desk and want a voice agent that can look things up and book things mid-call.

Worth knowing: Benchmark comparisons are Mistral's own evaluations. Voice cloning raises consent questions; only clone voices you have rights to.

Quick facts

Pricing	Mini Transcribe 3 $0.0035/min (batch), Realtime 3 $0.008/min, Small Transcribe 3 $0.0040/min, Mini TTS 3 $13/1M characters; voice agent API $0.04/min with Mistral Small 4 or $0.08/min with a larger model; on-prem by contacting sales
Free or paid	Paid (usage-based)
License	Proprietary API; self-deployment available (license terms not checked)not verified
Language	REST API / SDK; Model
Platform	Mistral EU-hosted API, or self-deployment on your own infrastructure or private cloud
GitHub stars	n/a (not a GitHub project)
Maker	Mistral AI

Related ranking

See how established tools compare in Best Voice & Text-to-Speech. New releases aren't scored until we review them against the rubric.

What we couldn't verify

quick_facts.license (self-deployment terms and whether weights are open)
highlights[0] (benchmark claims are Mistral's)

Checked 2026-10-22 from the model's own pages. Prices and star counts change, so confirm on the official site. Listing here is not a ranking or an endorsement.