Mistral Voxtral: open-weight speech understanding
Don’t want to install Mistral Voxtral? Drop the file here.
The first 20 minutes of any recording are transcribed free: a recording up to 20 minutes transcribes in full, a longer one is previewed, and a pack from $6 (500 minutes) transcribes the rest and opens every export, and credits never expire.
Voxtral Mini (3B) and Voxtral Small (24B) transcribe up to 30 minutes of audio, answer questions and summarise from speech, and call functions, under Apache-2.0 or through Mistral's API.
Get Mistral Voxtral Official site & pricing
Voxtral (Mini 3B, Small 24B) is Mistral's open speech model: 30-minute transcription, Q&A and summaries straight from audio, Apache-2.0. API from $0.001 a minute.
Used as a connector in self-hosted apps like Speakr. Over 30 minutes means splitting the file; a hosted service takes the whole recording.
What it is
Voxtral is Mistral AI's family of speech-understanding models, released in July 2025 in two open-weight sizes under Apache-2.0: Voxtral Mini (3B, built on Ministral 3B) for local and edge use, and Voxtral Small (24B) for production scale. With a 32k-token context they transcribe up to 30 minutes of audio, or 40 minutes for understanding tasks such as Q&A and summaries, detect the spoken language automatically, and can trigger function calls from speech. Mistral lists English, Spanish, French, Portuguese, Hindi, German, Dutch and Italian as its strongest languages. Mistral also serves Voxtral through its API, where transcription starts at $0.001 a minute per its announcement; vLLM is the recommended way to self-host.
Watch out for: Transcription up to 30 minutes per request (40 for understanding) within a 32k-token context; strongest in eight major languages. The 24B model needs serious GPU memory.
Recordings longer than 30 minutes, in one piece
- Drop the audio or video file into the box at the top of this page, or paste a link (YouTube, a podcast or RSS feed, Google Drive, Twitch or Kick). Nothing to install and no GPU needed.
- The language is detected automatically and the transcript stays in the language that was spoken (WhipScribe does not translate).
- Every job also returns a summary, chapters and the questions the recording answers, with a chat over the transcript.
- Read the summary free. A pack from $6 (500 minutes) opens the full transcript and the downloads: TXT, DOCX, SRT, VTT and JSON. Packs are one-time purchases and credits never expire.
- Files can be up to 10 hours or 5 GB per file.
Related: Bulk transcription · SRT generator · Transcribe large files · WhisperX · Pricing
Install / use
uv pip install -U "vllm[audio]" # model: mistralai/Voxtral-Mini-3B-2507Key specs
What people actually do with Mistral Voxtral-style transcription
The tool is the means. These are the jobs — each one priced at published rates, each one wired up on its own page.
Features
| Speaker diarization | Not stated |
| Word-level timestamps | Not stated |
| Streaming / real-time | Not stated |
| Languages supported | — |
| HIPAA eligible | No |
Links
- mistral.ai/news/voxtral ↗announcement: sizes, licence, API pricing
- mistralai/Voxtral-Mini-3B-2507 ↗3B weights · Apache-2.0
- mistralai/Voxtral-Small-24B-2507 ↗24B weights · Apache-2.0
Mistral Voxtral vs Whipscribe
| Feature | Mistral Voxtral | Whipscribe |
|---|---|---|
| Category | Open source | Transcription APIs |
| Pricing | free open weights (Apache-2.0); API transcription from $0.001 a minute | $6–$24 one-time packs (credits never expire) · first 20 minutes free |
| Speaker identification | No | Named in the summary, key quotes and chapters |
| Word timestamps | No | Yes |
| Streaming | No | No |
| Languages | — | 99 |
| Platforms | Linux, API | Web, API, MCP |
Alternatives to Mistral Voxtral
Frequently asked about Mistral Voxtral
Is Voxtral open source?
The weights of Voxtral Mini (3B) and Voxtral Small (24B) are released under Apache-2.0 on Hugging Face. Mistral also offers them through its paid API.
How long an audio file can Voxtral transcribe?
Up to 30 minutes for transcription and 40 minutes for understanding tasks, within its 32k-token context. Longer recordings must be split.
How much does the Voxtral API cost?
Mistral's announcement states API pricing starts at $0.001 a minute. Check Mistral's pricing page for current rates.
Which languages is Voxtral best at?
Mistral names English, Spanish, French, Portuguese, Hindi, German, Dutch and Italian. It detects the source language automatically.
How do I run Voxtral locally?
Mistral recommends vLLM (0.10.0 or later) with audio support; Transformers also works. Use temperature 0 for transcription.
Can I use Mistral Voxtral in the browser?
Mistral's announcement points to trying Voxtral in Le Chat, and developers can call it through Mistral's API. Self-hosting the open weights needs a GPU.
What is Mistral Voxtral?
Voxtral (Mini 3B, Small 24B) is Mistral's open speech model: 30-minute transcription, Q&A and summaries straight from audio, Apache-2.0. API from $0.001 a minute.
Is Mistral Voxtral HIPAA compliant?
Mistral Voxtral does not claim HIPAA compliance. For PHI, run it on infrastructure you control or choose a service that will sign a BAA.
What are the limitations of Mistral Voxtral?
Transcription up to 30 minutes per request (40 for understanding) within a 32k-token context; strongest in eight major languages. The 24B model needs serious GPU memory.
Who is Mistral Voxtral best for?
Developers who want one open model for transcription plus questions and summaries over audio, self-hosted with vLLM or through Mistral's API.
What should I know before choosing Mistral Voxtral?
Used as a connector in self-hosted apps like Speakr. Over 30 minutes means splitting the file; a hosted service takes the whole recording.
Who makes Mistral Voxtral?
Mistral Voxtral is made by Mistral AI.
What kind of tool is Mistral Voxtral?
In this directory Mistral Voxtral is filed under oss engine, audio llm as a open-source tool.
How is Whipscribe different from Mistral Voxtral?
Mistral Voxtral is something you run; Whipscribe is something you call. Recordings are transcribed on our own servers and never used for training, and every job returns five formats plus an AI summary, quotes and topics. The first 20 minutes of any recording are transcribed free, and a pack from $6 (500 minutes) transcribes the rest and opens every export.
What does transcription cost if I use Whipscribe instead of Mistral Voxtral?
$0.012 a minute — about $0.72 an audio hour — bought as credits that never expire. There is no subscription and no monthly minimum: a 3-hour recording is about $2.16, a thousand-hour archive is about $720 once. The first 20 minutes of any recording are transcribed free, and a pack from $6 (500 minutes) transcribes the rest and opens every export, so you can see what your recording holds before paying anything.
Whipscribe is a hosted transcription service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.
Transcribe your recordings
Running Mistral Voxtral yourself? Send the same file here and compare the transcript. Transcribe any language, any recording · Transcription services from $0.29 an hour · Transcribe from any meeting or recording platform · How to transcribe, step by step
Explore
All transcription tools · By language · By platform · By file format · How-to guides · Transcription services · Transcription API · Pricing
Have the recording already?
Skip the setup and send the file or the link: see transcribing a long MP3 or M4A, a podcast episode as text from its link and the SRT generator.
Rather have it done for you?
Free tools cost setup time. When you want a finished transcript instead: see an honest roundup of the best transcription services, affordable transcription priced per hour and how our transcription services work.