Swift Qwen3.8 27B MIX needs about 11.8 GB of memory at its main file (a 10.1 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 4 quants, from 0.0 GB to 14.2 GB.
Swift Qwen3.8 27B MIX – efficient reasoning with mixed‑tensor quantization
Swift Qwen3.8 27B MIX is a 27.3‑billion‑parameter model derived from Qwen3.8‑27B. It reduces thinking tokens by 58.3% while keeping performance loss below 1%, providing a 1.95× speed‑up on several tasks. The model is available in several mixed‑tensor quantizations designed for 16 GB VRAM GPUs.
Good at
- Fast reasoning with fewer tokens
- Near‑identical performance to dense Qwen3.8‑27B
- Optimized for 16 GB VRAM consumer GPUs
Limits
- No vision support
- Requires a 16 GB VRAM budget for best performance
- No second file needed – MTP head embedded
Best for
- Speed‑sensitive reasoning tasks
- Low‑latency inference on consumer GPUs
- Applications needing efficient token usage
Who it’s for. Developers and researchers who need a large‑scale reasoning model that runs efficiently on consumer GPUs. The model is suitable for building inference services where token efficiency and speed are critical.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 0.5 GB | Fits | Get file → | |
| — | 10.1 GB | 11.8 GB | Too big | Get file → | |
| — | 12.9 GB | 14.9 GB | Too big | Get file → | |
| — | 14.2 GB | 16.4 GB | Too big | Get file → |
For an 8 GB VRAM budget, use the 10.12 GB file which requires about 11.8 GB memory at 8 k context. For 12 GB VRAM, the 12.87 GB file uses about 14.9 GB memory. For 16 GB VRAM, the 14.16 GB file uses about 16.4 GB memory. The 0.01 GB file is a very small test version that uses only 0.5 GB memory.
What GPU do you need for Swift Qwen3.8 27B MIX?
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 the main file (0.0 GB file, about 0.5 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 0.5 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 0.5 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (10.1 GB file, about 11.8 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (12.9 GB file, about 14.9 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (14.2 GB file, about 16.4 GB in use).
Which Swift Qwen3.8 27B MIX quant should you choose?
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 Swift Qwen3.8 27B MIX
How much VRAM does Swift Qwen3.8 27B MIX need?
About 11.8 GB at its main file with an 8K context window. That is the 10.1 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 Swift Qwen3.8 27B MIX on an 8 GB GPU?
Yes, with the smaller files: the 0.0 GB file needs about 0.5 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Swift Qwen3.8 27B MIX GGUF?
It depends on the quant: file is 0.0 GB, file is 10.1 GB, file is 12.9 GB, file is 14.2 GB.
Where can I download Swift Qwen3.8 27B MIX GGUF?
The files are on Hugging Face in the vmarcelo/Swift-Qwen3.8-27B-MIX_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 Swift Qwen3.8 27B MIX 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 Swift Qwen3.8 27B MIX on an 8 GB VRAM GPU?
Yes, by using the 10.12 GB quantized file which requires about 11.8 GB memory at 8 k context.
What is the context length supported by this model?
The model is evaluated at an 8 k context length, which is the reference for memory estimates.
Does this model support vision tasks?
No, the model does not include vision capabilities.