Maq 1.0 4B SWE needs about 7.2 GB of memory at Q4_K_M (a 5.3 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 12 quants, from 4.4 GB (Q2_K) to 14.9 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.6 GB.
Maq 1.0 4B SWE – Gemma 4 E4B fine‑tuned for software engineering
Maq 1.0 4B SWE is a Gemma 4 E4B model post‑trained for coding, reasoning, tools and vision tasks. It was fine‑tuned with LoRA SFT on a mix of NVIDIA Nemotron code/SWE datasets and merged into full BF16 weights. The model supports a 131072 token context window.
Good at
- Coding assistance
- Tool‑using agentic traces
- Competitive programming
Limits
- No general chatbot use
- Not safety‑tuned
- No vision towers in SFT
Best for
- Local coding assistant
- SWE issue‑to‑patch conversations
- Agentic code generation
Who it’s for. Developers and researchers who need a lightweight, on‑device Gemma 4 E4B model fine‑tuned for software‑engineering tasks. The model is suitable for local experimentation and small‑scale coding assistant deployments.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 4.4 GB | 6.3 GB | Fits | Get file → | |
| Q3_K_S | 4.6 GB | 6.5 GB | Fits | Get file → | |
| Q3_K_M | 4.8 GB | 6.7 GB | Fits | Get file → | |
| Q3_K_L | 5.0 GB | 6.9 GB | Fits | Get file → | |
| IQ4_XS | 5.1 GB | 7.0 GB | Fits | Get file → | |
| Q4_K_S | 5.2 GB | 7.1 GB | Fits | Get file → | |
| Q4_K_M | 5.3 GB | 7.2 GB | Tight | Get file → | |
| Q5_K_S | 5.6 GB | 7.6 GB | Tight | Get file → | |
| Q5_K_M | 5.7 GB | 7.6 GB | Tight | Get file → | |
| Q6_K | 6.2 GB | 8.1 GB | Too big | Get file → | |
| Q8_0 | 8.0 GB | 9.9 GB | Too big | Get file → | |
| F16 | 14.9 GB | 16.9 GB | Too big | Get file → |
For 8 GB VRAM, Q4_K_M (5.3 GB) or Q5_K_M (5.72 GB) are recommended. If memory is limited to 6 GB, Q3_K_M (4.82 GB) or Q3_K_S (4.63 GB) provide a good balance. Higher‑precision Q8_0 (7.97 GB) or F16 (14.94 GB) require more VRAM and are suitable for high‑accuracy use cases.
What GPU do you need for Maq 1.0 4B SWE?
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: the best quant that fits is Q5_K_M (5.7 GB file, about 7.6 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (8.0 GB file, about 9.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (8.0 GB file, about 9.9 GB in use).
- 24 GB VRAM: the best quant that fits is F16 (14.9 GB file, about 16.9 GB in use).
Which Maq 1.0 4B SWE quant should you choose?
Q4_K_M is the usual starting point for Maq 1.0 4B SWE: it balances file size and quality well. Here it is a 5.3 GB file.
The Q8_0 file is 8.0 GB, about 1.5 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.
If memory is tight, the smallest file is Q2_K at 4.4 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 Maq 1.0 4B SWE
How much VRAM does Maq 1.0 4B SWE need?
About 7.2 GB at Q4_K_M with an 8K context window. That is the 5.3 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 Maq 1.0 4B SWE on an 8 GB GPU?
Yes, with Q5_K_M: the 5.7 GB file needs about 7.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Maq 1.0 4B SWE GGUF?
It depends on the quant: Q2_K is 4.4 GB, Q3_K_S is 4.6 GB, Q3_K_M is 4.8 GB, Q3_K_L is 5.0 GB, IQ4_XS is 5.1 GB, Q4_K_S is 5.2 GB and 6 more.
What context length does Maq 1.0 4B SWE support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.4 GB at 8K and 5.6 GB at 32K.
Where can I download Maq 1.0 4B SWE GGUF?
The files are on Hugging Face in the mradermacher/Maq-1.0-4B-SWE-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 Maq 1.0 4B SWE 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 Maq 1.0 4B SWE on 8 GB VRAM?
Yes, use the Q4_K_M or Q5_K_M quant, which fit within 8 GB VRAM and provide a balance of speed and quality.
Does this model support vision tasks?
No, the vision towers were dropped during fine‑tuning, so the model does not handle vision inputs.
What license applies to this model?
The model is released under the Apache‑2.0 license.