Qwen3.8 27B needs about 20.5 GB of memory at Q4_K_M (a 17.8 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 2 quants, from 15.4 GB (IQ4_XS) to 17.8 GB (Q4_K_M). Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.
Qwen3.8‑27B: 27B‑parameter dense multimodal model with 256K context
Qwen3.8‑27B is a 27.3‑billion‑parameter dense model that supports text, image, and video inputs. It offers a 262,144‑token native context and is available in two GGUF variants: multimodal Q4_K_M and text‑only IQ4_XS.
Good at
- Long‑horizon reasoning
- Multimodal understanding
- Tool‑calling in XML format
Limits
- Requires 24 GB+ VRAM for multimodal Q4_K_M
- Text‑only IQ4_XS is ~15 GB, still large for small GPUs
- No vision support in IQ4_XS
Best for
- Complex agentic tasks
- Research code generation
- Multimodal document analysis
Who it’s for. Developers and researchers needing a high‑capacity multimodal LLM that fits in a single GPU with aggressive quantization. Suitable for deployment on systems with 24 GB+ VRAM or with llama.cpp KV cache optimizations.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ4_XS | 15.4 GB | 18.1 GB | Too big | Get file → | |
| Q4_K_M | 17.8 GB | 20.5 GB | Too big | Get file → |
For 24 GB+ VRAM, use Q4_K_M (17.77 GB) to keep vision and video support. If limited to 16 GB VRAM, use IQ4_XS (15.42 GB) for text‑only workloads; memory usage is approximate and depends on context length.
What GPU do you need for Qwen3.8 27B?
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: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (17.8 GB file, about 20.5 GB in use).
Which Qwen3.8 27B quant should you choose?
Q4_K_M is the usual starting point for Qwen3.8 27B: it balances file size and quality well. Here it is a 17.8 GB file.
If memory is tight, the smallest file is IQ4_XS at 15.4 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Qwen3.8 27B
How much VRAM does Qwen3.8 27B need?
About 20.5 GB at Q4_K_M with an 8K context window. That is the 17.8 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 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 18.1 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Qwen3.8 27B GGUF?
It depends on the quant: IQ4_XS is 15.4 GB, Q4_K_M is 17.8 GB.
What context length does Qwen3.8 27B 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 Qwen3.8 27B GGUF?
The files are on Hugging Face in the impacte/Qwen3.8-27B-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 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 Qwen3.8‑27B on 8 GB VRAM?
No, both GGUF variants exceed 8 GB VRAM. The smallest variant requires 15.42 GB.
Does the IQ4_XS variant support images?
No, IQ4_XS is text‑only and removes the vision encoder.
What context length does the model support?
The model has a native context of 262,144 tokens (256K).