MistralVoxtral Small 2507
Mistral's Voxtral Small 24B 2507 is a speech-to-text model that excels at speech transcription, translation, and audio understanding, with a notable long-form context window of 131,072 tokens, allowing it to handle extended audio inputs. It features dedicated transcription mode, built-in Q&A and summarization, and natively multilingual capabilities, including automatic language detection.
Input
Output
Context
33K
Max Output
4K
Parameters
24.3B
Technical Specifications
Input Modalities
Output Modalities
Capabilities
Loading capabilities…
Estimated GPU requirements for Voxtral Small 2507
Estimates based on INT8 quantization at up to 32K context. A count above one assumes tensor parallelism across the cards. Actual requirements vary by framework and configuration.
More models by Mistral
The creator's other models in the catalog, with their context, size and license where published.
Compare it, or keep browsing
Put this model beside its alternatives on the same evidence, or go back to the full catalog.
Frequently asked questions about Voxtral Small 2507
Answered from the entry's own fields: context, license, modalities, evidence, serving and the memory to self-host.
Voxtral Small 2507 advertises a context window of 32,768 tokens, with a maximum output of 4,000 tokens in a single response. The figure is the creator's published maximum; a given provider may serve less, and the gateway routes on what each provider actually serves.
Yes. Voxtral Small 2507 is an open-weight model released under the Apache 2.0 license, so the weights can be downloaded and self-hosted within that license's terms.
Voxtral Small 2507 accepts Audio and Text and produces Text. The capabilities card on this page lists which API features each deployment honours, such as function calling and structured output, with the source each was checked against.
No published benchmark result for Voxtral Small 2507 is in the catalog yet, so the model is shown as unmeasured. It is not ranked or estimated; the router treats it as unknown for every task until a suite measures it.
Yes. Voxtral Small 2507 is served on the managed pool through the OpenAI-compatible endpoint as mistral-voxtral-small-2507, pinned by name or chosen by routing when it is the best fit for a request. The Try in Playground button opens it directly.
About 24.4 GB at INT8 for the weights and a default context, from the catalog's 24.3B parameter count; FP16 needs roughly twice that, and long contexts or many concurrent requests add KV cache on top. The GPU section on this page lists cards that hold it, and the capacity planner sizes it for your context length and traffic.
Fields collected from public registries, provider APIs and benchmark publishers, each tagged with its source.
Last updated: Nov 6, 2026
Put intelligence in the middle.
One gateway in front of every model, with your policies applied and every decision on record. Start on the free plan with 5,000 requests a month, no card required.