Speech to text · Mistral AI
Voxtral Mini 3B
Speech-understanding model: transcribe, then ask questions about the audio.
Datasheet
Maker | Mistral AI |
|---|---|
Parameters | 4.68 BVerifiedsafetensors.total from the HF API Source |
Licence | Apache-2.0Verified |
Access | Public download |
Task (HF pipeline) | — |
First published on HF | 2025-07-01 |
Last updated on HF | 2025-07-28 |
Weights in memory | 9.4 GB at 16-bit · 2.8 GB at ~4.85 bitsEstimateParameters × bits ÷ 8. Excludes context cache and runtime overhead; not every model has a 4-bit build. |
Benchmark
7.84%
Open ASR Leaderboard, average WER (English short-form) · #5 of 7 scored in this category
avg 7.8371, RTFx 179.78. Source
The description summarises the model card at https://huggingface.co/mistralai/Voxtral-Mini-3B-2507. Parameters and licence are read from the Hugging Face API; the weights-in-memory line is our arithmetic, not a measurement.