Wren needs about 78.9 GB of memory at Q4_K_M (a 77.6 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 8 quants, from 0.3 GB to 77.6 GB (Q4_K_M). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.
Compact Qwen4exp MoE model for consumer GPUs
Wren is a 117‑billion‑parameter Qwen4exp model with 50% of routed experts removed. It is designed for small, agile deployment on consumer GPUs, using a 12 GB VRAM target. The model is available in GGUF format with various quantisations.
Good at
- Efficient MoE inference on consumer GPUs
- Reduced expert count for lower memory
- Maintains reasoning speed of parent models
Limits
- Requires 12 GB VRAM for full performance
- No vision capabilities
- License requires attribution and may restrict commercial use
Best for
- Local chatbot deployment on laptops
- Rapid prototyping of Qwen4exp models
- Testing MoE pruning techniques
Who it’s for. Developers and researchers needing a lightweight Qwen4exp variant for local inference. It is suitable for those who can run llama.cpp on a 12 GB GPU and want to experiment with MoE pruning.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.3 GB | 1.6 GB | Fits | Get file → | |
| — | 0.3 GB | 1.6 GB | Fits | Get file → | |
| — | 58.7 GB | 60.0 GB | Too big | Get file → | |
| — | 60.4 GB | 61.7 GB | Too big | Get file → | |
| — | 60.4 GB | 61.7 GB | Too big | Get file → | |
| — | 60.4 GB | 61.7 GB | Too big | Get file → | |
| — | 62.7 GB | 64.0 GB | Too big | Get file → | |
| Q4_K_M | 77.6 GB | 78.9 GB | Too big | Get file → |
For 12 GB VRAM, use the Q4_K_M quantised file (77.62 GB, ~78.9 GB memory at 8 k context). If memory is tighter, the 60.38 GB files (~61.7 GB memory) are the next best option. Smaller 0.3 GB files are not full models and cannot be used alone.
What GPU do you need for Wren?
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 the main file (0.3 GB file, about 1.6 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
- 48 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 1.6 GB in use).
Which Wren quant should you choose?
Q4_K_M is the usual starting point for Wren: it balances file size and quality well. Here it is a 77.6 GB file.
If memory is tight, the smallest file is the first one in the table at 0.3 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Wren
How much VRAM does Wren need?
About 78.9 GB at Q4_K_M with an 8K context window. That is the 77.6 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 Wren on an 8 GB GPU?
Yes, with the smaller files: the 0.3 GB file needs about 1.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Wren GGUF?
It depends on the quant: file is 0.3 GB, file is 0.3 GB, file is 58.7 GB, file is 60.4 GB, file is 60.4 GB, file is 60.4 GB and 2 more.
What context length does Wren support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download Wren GGUF?
The files are on Hugging Face in the ohmysimo/Wren-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 Wren use?
Hugging Face lists the license as other. 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 Wren on 8 GB VRAM?
No, the smallest full model requires 12 GB VRAM for optimal performance.
Does Wren support vision tasks?
No, Wren is a text-only model with no vision capabilities.
What license applies to Wren?
Wren inherits Swift Open License v1.0 and Qwen Community License 1.0; commercial use above thresholds requires a separate license.