whisper medium needs about 1.1 GB of memory at Q4_1 (a 0.5 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 13 quants, from 0.3 GB (Q2_K) to 0.8 GB (Q8_0).
Whisper Medium quantized for efficient speech transcription
Whisper Medium is a transformer‑based encoder‑decoder model for automatic speech recognition and translation. The GGUF quantized versions reduce file size while preserving performance. The model is released under the MIT license.
Good at
- Speech transcription
- Speech translation
- Multilingual support
Limits
- No vision support
- Requires WhisperProcessor
- Context tokens needed
Best for
- Transcribing podcasts
- Translating live audio
- Batch audio processing
Who it’s for. Developers building speech‑to‑text or translation applications who need a lightweight, open‑source model. The quantized GGUF files allow deployment on devices with limited storage and memory.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 0.3 GB | 0.8 GB | Fits | Get file → | |
| Q3_K | 0.3 GB | 0.9 GB | Fits | Get file → | |
| Q4_0 | 0.4 GB | 1.0 GB | Fits | Get file → | |
| Q4_0 | 0.4 GB | 1.0 GB | Fits | Get file → | |
| Q4_K | 0.4 GB | 1.0 GB | Fits | Get file → | |
| Q4_1 | 0.5 GB | 1.0 GB | Fits | Get file → | |
| Q4_1 | 0.5 GB | 1.1 GB | Fits | Get file → | |
| Q5_0 | 0.5 GB | 1.1 GB | Fits | Get file → | |
| Q5_K | 0.5 GB | 1.1 GB | Fits | Get file → | |
| Q5_1 | 0.6 GB | 1.2 GB | Fits | Get file → | |
| Q6_K | 0.6 GB | 1.2 GB | Fits | Get file → | |
| Q8_0 | 0.8 GB | 1.4 GB | Fits | Get file → | |
| Q8_0 | 0.8 GB | 1.4 GB | Fits | Get file → |
For 8 GB VRAM, Q4_K (0.44 GB) or Q5_0 (0.54 GB) are suitable. For 4 GB VRAM, Q2_K (0.27 GB) or Q3_K (0.34 GB) fit. Memory estimates are approximate and depend on context length.
What GPU do you need for whisper medium?
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 Q8_0 (0.8 GB file, about 1.4 GB in use).
Which whisper medium quant should you choose?
Q4_1 (0.5 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is Q2_K at 0.3 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about whisper medium
How much VRAM does whisper medium need?
About 1.1 GB at Q4_1 with an 8K context window. That is the 0.5 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 whisper medium on an 8 GB GPU?
Yes, with Q8_0: the 0.8 GB file needs about 1.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of whisper medium GGUF?
It depends on the quant: Q2_K is 0.3 GB, Q3_K is 0.3 GB, Q4_0 is 0.4 GB, Q4_0 is 0.4 GB, Q4_K is 0.4 GB, Q4_1 is 0.5 GB and 7 more.
Where can I download whisper medium GGUF?
The files are on Hugging Face in the oxide-lab/whisper-medium-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 whisper medium use?
Hugging Face lists the license as mit. 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 Whisper Medium on 8 GB VRAM?
Yes, using Q4_K or Q5_0 quantizations which require about 1 GB of VRAM at 8k context.
Does this model support speech translation?
Yes, Whisper Medium can translate audio to another language when the appropriate context tokens are provided.
What license does the model use?
The model is released under the MIT license, identical to the original Whisper.