s1-mini needs about 2.0 GB of memory at Q4_K_M (a 0.5 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 35 quants, from 0.0 GB to 1.5 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.8 GB.
s1-mini: 0.6B‑parameter text normalizer for speech‑to‑text output
s1-mini is a 0.6B‑parameter model fine‑tuned from Qwen/Qwen3‑0.6B. It rewrites raw ASR transcripts into clean written text, removing fillers, resolving self‑corrections, and adding punctuation and capitalization. The model is designed for single‑purpose text normalisation and is not a general chat model.
Good at
- ASR transcript normalisation
- English‑only text cleaning
- Low‑memory inference
Limits
- Not a chat model
- Requires a control line in input
- English only
Best for
- Dictation apps
- Speech‑to‑text pipelines
- Post‑processing ASR output
Who it’s for. Developers building speech‑to‑text systems who need accurate, clean text output. The model is lightweight and runs comfortably on laptop CPUs.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 1.4 GB | Fits | Get file → | |
| IQ3_XXS | 0.4 GB | 1.8 GB | Fits | Get file → | |
| IQ3_XS | 0.4 GB | 1.8 GB | Fits | Get file → | |
| IQ3_S | 0.4 GB | 1.9 GB | Fits | Get file → | |
| IQ3_M | 0.4 GB | 1.9 GB | Fits | Get file → | |
| Q4_0 | 0.4 GB | 1.9 GB | Fits | Get file → | |
| Q3_K_S | 0.4 GB | 1.9 GB | Fits | Get file → | |
| IQ4_XS | 0.5 GB | 1.9 GB | Fits | Get file → | |
| IQ4_NL | 0.5 GB | 1.9 GB | Fits | Get file → | |
| Q4_1 | 0.5 GB | 1.9 GB | Fits | Get file → | |
| Q3_K_M | 0.5 GB | 1.9 GB | Fits | Get file → | |
| Q4_K_S | 0.5 GB | 2.0 GB | Fits | Get file → | |
| Q5_0 | 0.5 GB | 2.0 GB | Fits | Get file → | |
| Q4_K_M | 0.5 GB | 2.0 GB | Fits | Get file → | |
| Q3_K_L | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q5_1 | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q5_K_S | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q5_K_M | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q4_0 | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q4_K_L | 0.6 GB | 2.0 GB | Fits | Get file → | |
| Q4_1 | 0.6 GB | 2.1 GB | Fits | Get file → | |
| Q6_K_M | 0.6 GB | 2.1 GB | Fits | Get file → | |
| Q5_0 | 0.6 GB | 2.1 GB | Fits | Get file → | |
| Q5_K_L | 0.7 GB | 2.1 GB | Fits | Get file → | |
| Q5_1 | 0.7 GB | 2.1 GB | Fits | Get file → | |
| Q6_K_L | 0.7 GB | 2.1 GB | Fits | Get file → | |
| Q8_0 | 0.8 GB | 2.2 GB | Fits | Get file → | |
| BF16 | 1.1 GB | 2.6 GB | Fits | Get file → | |
| F16 | 1.1 GB | 2.6 GB | Fits | Get file → | |
| BF16 | 1.2 GB | 2.6 GB | Fits | Get file → | |
| F16 | 1.2 GB | 2.6 GB | Fits | Get file → | |
| BF16 | 1.3 GB | 2.7 GB | Fits | Get file → | |
| F16 | 1.3 GB | 2.7 GB | Fits | Get file → | |
| BF16 | 1.5 GB | 2.9 GB | Fits | Get file → | |
| F16 | 1.5 GB | 2.9 GB | Fits | Get file → |
For 8 GB VRAM, use IQ3_XXS (0.39 GB) or IQ3_XS (0.40 GB). For 4 GB VRAM, IQ3_XXS is recommended. If more VRAM is available, Q4_0 (0.43 GB) or Q4_K_S (0.52 GB) provide higher quality with still modest memory use. Estimates are approximate.
What GPU do you need for s1-mini?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) 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 BF16 (1.5 GB file, about 2.9 GB in use).
Which s1-mini quant should you choose?
Q4_K_M is the usual starting point for s1-mini: it balances file size and quality well. Here it is a 0.5 GB file.
The Q8_0 file is 0.8 GB, about 1.5 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.
If memory is tight, the smallest file is the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about s1-mini
How much VRAM does s1-mini need?
About 2.0 GB at Q4_K_M 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 s1-mini on an 8 GB GPU?
Yes, with BF16: the 1.5 GB file needs about 2.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of s1-mini GGUF?
It depends on the quant: file is 0.0 GB, IQ3_XXS is 0.4 GB, IQ3_XS is 0.4 GB, IQ3_S is 0.4 GB, IQ3_M is 0.4 GB, Q4_0 is 0.4 GB and 29 more.
What context length does s1-mini support?
The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 0.9 GB at 8K and 3.8 GB at 32K.
Where can I download s1-mini GGUF?
The files are on Hugging Face in the Mungert/s1-mini-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 s1-mini use?
Hugging Face lists the license as other. 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 s1-mini on 4 GB VRAM?
Yes, the IQ3_XXS quantised model is 0.39 GB and runs comfortably on 4 GB VRAM.
Does s1-mini support multiple languages?
No, the current release supports English only.
Is s1-mini a chat model?
No, it is a single‑purpose text normaliser and will not follow general instructions.