Qwen3.8 Perian needs about 93.5 GB of memory at Q6_K_XL (a 83.1 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 5 quants, from 47.5 GB (IQ4_XS) to 155.8 GB (BF16).
Qwen3.8 Perian – 77.5B parameter Qwen model compressed with MTP and PLE reductions
Qwen3.8 Perian is a 77.5B parameter Qwen model that has undergone depth pruning, routed‑expert width reduction, and a 50% PLE n‑gram table reduction. The GGUF release includes embedded MTP depth 3 and supports a 245,760‑token context. It is licensed under apache‑2.0.
Good at
- Large‑context text generation
- Document‑disjoint reasoning
- Embedded MTP action planning
Limits
- No vision support
- Requires transformers_compact_runtime.py
- Not a production‑ready release
Best for
- Research on long‑context LLMs
- Testing MTP behavior
- Fine‑tuning compressed Qwen models
Who it’s for. Researchers and developers who need a compressed Qwen3.8 model with long‑context support and embedded MTP. The model is suitable for experimentation rather than production deployment.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ4_XS | 47.5 GB | 53.7 GB | Too big | Get file → | |
| Q5_K_XL | 73.4 GB | 82.7 GB | Too big | Get file → | |
| Q6_K_XL | 83.1 GB | 93.5 GB | Too big | Get file → | |
| Q8_0 | 83.4 GB (2 parts) | 94.0 GB | Too big | Get file → | |
| BF16 | 155.8 GB (5 parts) | 175.0 GB | Too big | Get file → |
For 8 GB VRAM, use IQ4_XS (47.5 GB file, ~53.7 GB memory at 8k context). For 16 GB VRAM, Q5_K_XL (73.39 GB file, ~82.7 GB memory) or Q6_K_XL (83.05 GB file, ~93.5 GB memory). BF16 and Q8_0 require more than 32 GB VRAM.
What GPU do you need for Qwen3.8 Perian?
These figures add the file size, an fp16 KV cache at an 8K context window 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: no quant fits at an 8K context.
- 32 GB VRAM: no quant fits at an 8K context.
- 48 GB VRAM: no quant fits at an 8K context.
Which Qwen3.8 Perian quant should you choose?
Q6_K_XL (83.1 GB) is the middle option in this repository. Very close to the original quality for most uses, at a clearly larger size than 4-bit.
If memory is tight, the smallest file is IQ4_XS at 47.5 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 Perian
How much VRAM does Qwen3.8 Perian need?
About 93.5 GB at Q6_K_XL with an 8K context window. That is the 83.1 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 Perian on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 53.7 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Qwen3.8 Perian GGUF?
It depends on the quant: IQ4_XS is 47.5 GB, Q5_K_XL is 73.4 GB, Q6_K_XL is 83.1 GB, Q8_0 is 83.4 GB, BF16 is 155.8 GB.
What context length does Qwen3.8 Perian support?
The GGUF header lists a trained context length of 262,144 tokens (256K).
Where can I download Qwen3.8 Perian GGUF?
The files are on Hugging Face in the jakeatx/Qwen3.8-Perian-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 Perian 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 Perian on 8 GB VRAM?
Use the IQ4_XS quantized file; it requires approximately 53.7 GB memory at 8k context, so 8 GB VRAM is insufficient.
Does this model support vision tasks?
No, the model has no vision capability.
What license applies to this model?
The model is released under the apache‑2.0 license.