QwenGram-4B REAL-15M + linear750 needs about 4.5 GB of memory at Q4_K_M (a 2.9 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 4 quants, from 2.9 GB (Q4_K_M) to 8.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.4 GB.
QwenGram-4B REAL-15M + linear750 – a 4.35B parameter Qwen3.5 derivative
QwenGram-4B REAL-15M + linear750 is a frozen Qwen3.5-4B backbone with a 15M reader and linear arbitration. It supports reasoning and tool usage. The model is released under apache-2.0 and requires an external PLE sidecar.
Good at
- reasoning
- tools
- post‑training reader optimization
Limits
- no vision support
- requires external PLE
- context length not specified
Best for
- small‑to‑medium inference on CPUs
- tool‑enabled chatbots
- post‑training fine‑tuning
Who it’s for. Developers needing a lightweight Qwen3.5 variant with reader augmentation. The model is suitable for applications that can supply the required PLE and benefit from the 15M reader tokens.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 2.9 GB | 4.5 GB | Fits | Get file → | |
| Q6_K | 3.7 GB | 5.3 GB | Fits | Get file → | |
| Q8_0 | 4.7 GB | 6.3 GB | Fits | Get file → | |
| BF16 | 8.8 GB | 10.4 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (file size 2.89 GB, estimated 4.5 GB memory). For 12 GB VRAM, Q6_K (3.67 GB, 5.3 GB memory) is appropriate. BF16 requires 10.4 GB memory and 8.77 GB file size.
What GPU do you need for QwenGram-4B REAL-15M + linear750?
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: the best quant that fits is Q6_K (3.7 GB file, about 5.3 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (4.7 GB file, about 6.3 GB in use).
- 12 GB VRAM: the best quant that fits is BF16 (8.8 GB file, about 10.4 GB in use).
Which QwenGram-4B REAL-15M + linear750 quant should you choose?
Q4_K_M is the usual starting point for QwenGram-4B REAL-15M + linear750: it balances file size and quality well. Here it is a 2.9 GB file.
The Q8_0 file is 4.7 GB, about 1.6 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about QwenGram-4B REAL-15M + linear750
How much VRAM does QwenGram-4B REAL-15M + linear750 need?
About 4.5 GB at Q4_K_M with an 8K context window. That is the 2.9 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 QwenGram-4B REAL-15M + linear750 on an 8 GB GPU?
Yes, with Q8_0: the 4.7 GB file needs about 6.3 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of QwenGram-4B REAL-15M + linear750 GGUF?
It depends on the quant: Q4_K_M is 2.9 GB, Q6_K is 3.7 GB, Q8_0 is 4.7 GB, BF16 is 8.8 GB.
What context length does QwenGram-4B REAL-15M + linear750 support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.4 GB at 32K.
Where can I download QwenGram-4B REAL-15M + linear750 GGUF?
The files are on Hugging Face in the Ninnix96/Qwengram-4B repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does QwenGram-4B REAL-15M + linear750 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 QwenGram-4B on 8 GB VRAM?
Yes, the Q4_K_M quantization uses about 4.5 GB memory at 8 k context, fitting within 8 GB VRAM.
Do I need a GPU to run this model?
No, the model can be run on CPU with llama.cpp, but a GPU accelerates inference.
What external files are required?
An external PLE sidecar is required; the PLE is approximately 32 GB and does not need to be loaded into GPU memory.