Swift Qwen3.8 27B Uncensored Dynamic MTP needs about 18.0 GB of memory at Q4_K_S (a 15.4 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 12 quants, from 9.8 GB (Q2_K_XL) to 54.7 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.
Swift Qwen3.8 27B Uncensored Dynamic MTP – a vision‑enabled, uncensored LLM
Swift Qwen3.8 27B Uncensored Dynamic MTP is a 0.461 billion‑parameter Qwen model with a 262144‑token context. It includes a vision tower and an MTP head for self‑speculative decoding. The model is released under an other license and is quantized for llama.cpp.
Good at
- Vision input and video processing
- Uncensored text generation
- Self‑speculative decoding via MTP
Limits
- No explicit language support listed
- Requires 8 k context for memory estimates
- License is other, not open source
Best for
- Developers needing vision‑enabled LLMs
- Applications requiring uncensored responses
- Environments with limited GPU memory
Who it’s for. Developers and researchers who need a vision‑capable, uncensored LLM that can run on llama.cpp. The model is suitable for projects that require self‑speculative decoding and can accommodate the memory requirements of the chosen quantization.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K_XL | 9.8 GB | 12.5 GB | Too big | Get file → | |
| IQ3_XXS | 10.9 GB | 13.6 GB | Too big | Get file → | |
| IQ3_S | 12.0 GB | 14.7 GB | Too big | Get file → | |
| Q3_K_XL | 13.1 GB | 15.8 GB | Too big | Get file → | |
| IQ4_XS | 14.3 GB | 16.9 GB | Too big | Get file → | |
| Q4_K_S | 15.4 GB | 18.0 GB | Too big | Get file → | |
| Q4_K_XL | 17.6 GB | 20.2 GB | Too big | Get file → | |
| Q5_K_S | 18.7 GB | 21.3 GB | Too big | Get file → | |
| Q5_K_M | 19.8 GB | 22.5 GB | Too big | Get file → | |
| Q6_K_XL | 25.3 GB | 28.0 GB | Too big | Get file → | |
| Q8_K_XL | 31.5 GB | 34.1 GB | Too big | Get file → | |
| BF16 | 54.7 GB | 57.3 GB | Too big | Get file → |
For 8 GB VRAM, Q4_K_XL (17.56 GB file, 20.2 GB memory at 8 k context) is a good balance. If 12 GB VRAM is available, Q5_K_M (19.77 GB file, 22.5 GB memory) offers higher quality. For 16 GB VRAM, Q6_K_XL (25.3 GB file, 28 GB memory) provides a larger model with acceptable performance.
What GPU do you need for Swift Qwen3.8 27B Uncensored Dynamic MTP?
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: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: the best quant that fits is Q3_K_XL (13.1 GB file, about 15.8 GB in use).
- 24 GB VRAM: the best quant that fits is Q5_K_M (19.8 GB file, about 22.5 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K_XL (25.3 GB file, about 28.0 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_K_XL (31.5 GB file, about 34.1 GB in use).
Which Swift Qwen3.8 27B Uncensored Dynamic MTP quant should you choose?
Q4_K_S (15.4 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_XL at 9.8 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 Swift Qwen3.8 27B Uncensored Dynamic MTP
How much VRAM does Swift Qwen3.8 27B Uncensored Dynamic MTP need?
About 18.0 GB at Q4_K_S with an 8K context window. That is the 15.4 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 Uncensored Dynamic MTP on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 12.5 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Swift Qwen3.8 27B Uncensored Dynamic MTP GGUF?
It depends on the quant: Q2_K_XL is 9.8 GB, IQ3_XXS is 10.9 GB, IQ3_S is 12.0 GB, Q3_K_XL is 13.1 GB, IQ4_XS is 14.3 GB, Q4_K_S is 15.4 GB and 6 more.
What context length does Swift Qwen3.8 27B Uncensored Dynamic MTP support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.2 GB at 8K and 8.7 GB at 32K.
Where can I download Swift Qwen3.8 27B Uncensored Dynamic MTP GGUF?
The files are on Hugging Face in the ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-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 Uncensored Dynamic MTP 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 on 8 GB VRAM?
Q4_K_XL requires about 20.2 GB memory at 8 k context, so 8 GB VRAM is insufficient. Use a smaller quant or reduce context length.
Does the model support image input?
Yes, the vision tower is included and supports image and video input through the projector.
Is the model open source?
The license is listed as other; it is not explicitly open source and may have restrictions.