PAW 27B X3 needs about 8.5 GB of memory at its main file (a 7.8 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 6 quants, from 0.7 GB (Q2_K) to 12.3 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 0.7 GB.
PAW 27B X3 – 4.85B parameter model with 256k context, 24 GB GPU friendly
PAW 27B X3 is a 4.85 B parameter model derived from Qwen3.8‑27B. It supports a 262144 token context and is optimized for a single 24 GB GPU using the PAW x3 trellis format. The model is released under the Apache‑2.0 license.
Good at
- Large context handling
- Efficient 1.5‑3.5 bpw quantization
- Runs on a single 24 GB GPU
Limits
- Requires llama‑paw fork, not standard llama.cpp
- No vision support in this variant
- No explicit language support listed
Best for
- Long‑form text generation
- Complex reasoning tasks
- Large‑context data processing
Who it’s for. Developers needing a dense, high‑context model that fits on a single 24 GB GPU. Suitable for research or production inference where the PAW x3 format is supported.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 0.7 GB | 1.4 GB | Fits | Get file → | |
| — | 6.3 GB | 6.9 GB | Fits | Get file → | |
| — | 7.8 GB | 8.5 GB | Too big | Get file → | |
| — | 9.3 GB | 10.0 GB | Too big | Get file → | |
| — | 10.8 GB | 11.5 GB | Too big | Get file → | |
| — | 12.3 GB | 13.0 GB | Too big | Get file → |
For 8 GB VRAM, use the Q2_K file (0.71 GB, ~1.4 GB memory at 8 k context). For 24 GB VRAM, any of the larger quant files will fit; the 6.27 GB file uses ~6.9 GB memory, while the 12.35 GB file uses ~13 GB memory. Choose the smallest file that meets your accuracy needs.
What GPU do you need for PAW 27B X3?
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: the best quant that fits is Q2_K (0.7 GB file, about 1.4 GB in use).
- 6 GB VRAM: the best quant that fits is Q2_K (0.7 GB file, about 1.4 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (6.3 GB file, about 6.9 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (10.8 GB file, about 11.5 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (12.3 GB file, about 13.0 GB in use).
Which PAW 27B X3 quant should you choose?
If memory is tight, the smallest file is Q2_K at 0.7 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 PAW 27B X3
How much VRAM does PAW 27B X3 need?
About 8.5 GB at its main file with an 8K context window. That is the 7.8 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 PAW 27B X3 on an 8 GB GPU?
Yes, with the smaller files: the 6.3 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of PAW 27B X3 GGUF?
It depends on the quant: Q2_K is 0.7 GB, file is 6.3 GB, file is 7.8 GB, file is 9.3 GB, file is 10.8 GB, file is 12.3 GB.
What context length does PAW 27B X3 support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.7 GB at 32K.
Where can I download PAW 27B X3 GGUF?
The files are on Hugging Face in the lackonendes/PAW-27B-X3-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 PAW 27B X3 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 PAW 27B X3 on a 24 GB GPU?
Yes, the model is designed to run on a single 24 GB GPU using the PAW x3 format.
Does PAW 27B X3 support vision tasks?
No, this variant does not include vision capabilities.