Voxtral Mini 3B needs about 3.8 GB of memory at Q4_K_M (a 3.0 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 6 quants, from 3.0 GB (Q4_K_M) to 9.4 GB (F16).
Voxtral Mini 3B – a 4.7B parameter audio‑LLM for transcription and translation
Voxtral Mini 3B is a 4.7 billion‑parameter model based on the Mistral architecture. It adds a Whisper‑large‑v3 audio encoder to the Mistral‑3B backbone, enabling speech transcription, translation, and audio understanding. The model supports a 32k‑token context length and multilingual operation.
Good at
- Speech transcription with low WER
- Multilingual audio translation
- Voice‑driven function calling
Limits
- No vision support
- Requires 8 k context for memory estimates
- Apache‑2.0 license
Best for
- Transcribing long audio files
- Building voice‑controlled applications
- Generating audio summaries
Who it’s for. Developers building audio‑centric LLM services or voice assistants. The model is suitable for teams that need a single model for both text and audio tasks and can run it locally with GGUF support.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 3.0 GB | 3.8 GB | Fits | Get file → | |
| Q5_K_M | 3.5 GB | 4.4 GB | Fits | Get file → | |
| Q6_K | 3.9 GB | 4.8 GB | Fits | Get file → | |
| Q8_0 | 5.0 GB | 6.1 GB | Fits | Get file → | |
| BF16 | 9.4 GB | 11.0 GB | Too big | Get file → | |
| F16 | 9.4 GB | 11.0 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (3.8 GB at 8k context). If 12 GB VRAM is available, Q5_K_M (4.4 GB) offers better precision. For higher accuracy and more VRAM, Q8_0 (6.1 GB) or BF16 (11 GB) can be used.
What GPU do you need for Voxtral Mini 3B?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is Q4_K_M (3.0 GB file, about 3.8 GB in use).
- 6 GB VRAM: the best quant that fits is Q6_K (3.9 GB file, about 4.8 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (5.0 GB file, about 6.1 GB in use).
- 12 GB VRAM: the best quant that fits is F16 (9.4 GB file, about 11.0 GB in use).
Which Voxtral Mini 3B quant should you choose?
Q4_K_M is the usual starting point for Voxtral Mini 3B: it balances file size and quality well. Here it is a 3.0 GB file.
The Q8_0 file is 5.0 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Voxtral Mini 3B
How much VRAM does Voxtral Mini 3B need?
About 3.8 GB at Q4_K_M with an 8K context window. That is the 3.0 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Voxtral Mini 3B on an 8 GB GPU?
Yes, with Q8_0: the 5.0 GB file needs about 6.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Voxtral Mini 3B GGUF?
It depends on the quant: Q4_K_M is 3.0 GB, Q5_K_M is 3.5 GB, Q6_K is 3.9 GB, Q8_0 is 5.0 GB, BF16 is 9.4 GB, F16 is 9.4 GB.
Where can I download Voxtral Mini 3B GGUF?
The files are on Hugging Face in the handy-computer/Voxtral-Mini-3B-2507-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Voxtral Mini 3B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Voxtral Mini 3B on 8 GB VRAM?
Yes, the Q4_K_M quantization requires about 3.8 GB of VRAM at 8k context, which fits within 8 GB.
Does the model support image input?
No, Voxtral Mini 3B does not have vision capabilities.
What context length does the model support?
The model supports a 32k‑token context length for audio and text tasks.