grug 27b qat q4 needs about 19.2 GB of memory at Q4_K_M (a 16.5 GB file) with an 8K context window, so it fits on a 24 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
27B Qwen3.5 model with QAT‑trained Q4_K_M quantization
grug‑27b is a 27‑billion‑parameter Qwen3.5 model fine‑tuned with LoRA and QAT. It is released under Apache‑2.0 and supports reasoning and vision. The Q4_K_M quantized GGUF file is 16.55 GB and requires about 19.2 GB of VRAM at 8 k context.
Good at
- Efficient reasoning with low token usage
- Strong coding and math performance
- Stable tool‑handing in agent settings
Limits
- No vision support in llama.cpp builds
- Requires recent llama.cpp with qwen3_5 support
- Large context (262 144 tokens) may need more VRAM
Best for
- Short‑form reasoning tasks
- Coding and math problem solving
- Agent‑based tool interactions
Who it’s for. Developers and researchers needing a large Qwen3.5 model with efficient Q4_K_M quantization. The model is suitable for local deployment on GPUs with at least 20 GB VRAM and recent llama.cpp builds.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 16.5 GB | 19.2 GB | Too big | Get file → |
For 8 GB VRAM use a higher‑bit quant like Q8_K_M if available. With 16 GB VRAM the Q4_K_M file (16.55 GB) can be loaded, but expect around 19.2 GB VRAM usage at 8 k context. For 32 GB VRAM the Q4_K_M quantization is optimal and provides the best balance of speed and quality.
What GPU do you need for grug 27b qat q4?
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 (16.5 GB file, about 19.2 GB in use).
Which grug 27b qat q4 quant should you choose?
Q4_K_M is the usual starting point for grug 27b qat q4: it balances file size and quality well. Here it is a 16.5 GB file.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about grug 27b qat q4
How much VRAM does grug 27b qat q4 need?
About 19.2 GB at Q4_K_M with an 8K context window. That is the 16.5 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 grug 27b qat q4 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 19.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of grug 27b qat q4 GGUF?
It depends on the quant: Q4_K_M is 16.5 GB.
What context length does grug 27b qat q4 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 grug 27b qat q4 GGUF?
The files are on Hugging Face in the ProCreations/grug-27b-qat-q4-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 grug 27b qat q4 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 grug‑27b on 8 GB VRAM?
No, the Q4_K_M file requires about 19.2 GB VRAM at 8 k context. A higher‑bit quant would be needed for 8 GB.
Does this model support vision?
The model has vision detection but llama.cpp builds currently do not support vision for this architecture.
What license applies to grug‑27b?
The model is released under the Apache‑2.0 license.