mistralai
Mistral: Voxtral Mini 3B 2507
mistralai/voxtral-mini-3b-2507
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Voxtral Mini 3B 2507 is Mistral AI's small audio understanding model, derived from Ministral 3B, that accepts audio input and produces transcription output; its Hugging Face card additionally describes audio Q&A, summarization, translation and voice-triggered function calling. It offers automatic language detection across eight languages and a 32k token context covering roughly 30 minutes of transcription audio. Evidence comes from Mistral's model card and the OpenRouter catalog; benchmark figures appear only as images without extractable values, so real-world accuracy should be validated on your own workloads.[1][2]
Strengths
- Mistral's model card describes a dedicated transcription mode with automatic language detection across English, French, German, Spanish, Italian, Portuguese, Dutch and Hindi, plus speech translation and audio understanding.[2]
- A 32k token context reportedly handles around 30 minutes of audio for transcription or 40 minutes for understanding, and built-in audio Q&A, summarization and voice-triggered function calling are documented.[2]
- The card reports roughly 9.5 GB of GPU memory in bf16/fp16, Apache 2.0 licensing, and support in vLLM and Transformers, which may ease self-hosting.[2]
Limitations
- System prompts are not yet supported, per the model card.[2]
- Benchmark results are published only as images with no extractable values, and no independent evaluation appears in this source pack; accuracy on specific workloads is unverified.[1][2]
- The OpenRouter catalog exposes the model only as audio-to-transcription, so the chat-style audio features described by Mistral may not be available through that provider.[1][2]
Where to use it
- Transcription and translation of multilingual meetings, calls or media in the listed languages, after validating accuracy on your audio conditions.[1][2]
- Voice interfaces that answer questions, summarize recordings or trigger backend functions from speech, if the deployment path supports those features.[2]
- Self-hosted audio pipelines where a roughly 3B-parameter model fits available GPU memory better than larger speech systems.[2]
Sources
- [1] OpenRouter catalog · Mistral: Voxtral Mini 3B 2507Sources checked: Nov 13, 2026
- [2] Model documentation · Mistral: Voxtral Mini 3B 2507Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Voxtral Mini 3B 2507 is a speech and audio understanding model from Mistral AI. It is suited for transcription, translation, and compact audio processing workloads.
Specifications
- Model ID
- mistralai/voxtral-mini-3b-2507
- Canonical ID
- mistralai/voxtral-mini-3b-2507-20260813
- Creator
- mistralai
- Added to OpenRouter
- 13.08.2026
- Context length
- 0
- Modality
- audio->transcription
- Input modalities
- audio
- Output modalities
- transcription
- Token pricing
- No per-token price in the source snapshot — see OpenRouter for the billing unit
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | — |
| Coding | No measurement | — |
| Agents | No measurement | — |
| Writing | No measurement | — |
| Reasoning | No measurement | — |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | input: audio |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
No evaluations are supplied for this model.
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| DeepInfrabf16 | 0 | — | — | — | — | — |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.