Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS needs about 12.2 GB of memory at its main file (a 9.6 GB file) with an 8K context window, so it fits on a 16 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
Qwen3.8‑27B LowGPU uncensored NoMTP IQ3XXXS – 12 GB VRAM tuned GGUF
This model is a 27 B parameter Qwen variant quantized for a 12 GB VRAM target. It removes MTP and uses a custom mixed‑precision recipe. The GGUF file is 9.57 GB and requires about 12.2 GB of VRAM at 8 k context.
Good at
- Large context 262 144 tokens
- Mixed‑precision per‑tensor quantization
- Uncensored reasoning and vision
Limits
- Requires ~12 GB VRAM
- No MTP support
- Uncensored – safety guardrails removed
Best for
- Deploying on RTX 4070 12 GB GPUs
- Running full GPU offload with 64 k context
- Using llama.cpp server with Hermes Agent
Who it’s for. Developers needing a 27 B Qwen model that fits a 12 GB GPU and can handle vision tasks. It is suitable for research or production where uncensored outputs are acceptable.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 9.6 GB | 12.2 GB | Too big | Get file → |
The repository provides a single GGUF quantization. The file size is 9.57 GB and it estimates 12.2 GB of VRAM at 8 k context. Use this quant if you have a 12 GB GPU; no other quants are available.
What GPU do you need for Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS?
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 the main file (9.6 GB file, about 12.2 GB in use).
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS
How much VRAM does Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS need?
About 12.2 GB at its main file with an 8K context window. That is the 9.6 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 Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 12.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS GGUF?
It depends on the quant: file is 9.6 GB.
What context length does Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.6 GB at 32K.
Where can I download Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS GGUF?
The files are on Hugging Face in the TheWegemann/Qwen3.8-27B-LowGPU-uncensored-NoMTP-IQ3XXXS-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 Qwen3.8 27B LowGPU uncensored NoMTP IQ3XXXS 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 this model on an 8 GB GPU?
No. The model requires about 12.2 GB of VRAM at 8 k context, so an 8 GB GPU is insufficient.
Does this model support vision?
Yes, it has detected vision ability.
Is the model safe for general use?
No. It is an uncensored, abliterated variant with reduced safety guardrails.