Voxtral Small 24B
Open 24B audio-language model for transcription and audio QA.
About
Voxtral Small 24B is Mistral AI's first open-weight speech model, released in July 2025 under Apache 2.0. It extends Mistral Small 3 with audio input, covering speech transcription, translation, audio question answering, and summarization inside a 32K context window, with function calling and structured outputs supported.
Mistral's launch benchmarks report it beating Whisper across every FLEURS task, and the 2507 checkpoint refresh followed in October 2025. A 3B Mini sibling targets edge and local deployment at about 9.5GB in bf16.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 125; Trending score 82/100 from researcher assessment.