Assistant needs about 1.8 GB of memory at Q4_0 (a 0.3 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 6 quants, from 0.2 GB (IQ3_M) to 0.9 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.
Gemma 4 26B A4B Assistant GGUF – instruction‑tuned, 0.42B parameters
Gemma 4 26B A4B Assistant is a 0.42 billion‑parameter instruction‑tuned model. It supports up to 131 072 context tokens and is released under the MIT license. The model is available in several GGUF quantizations for different memory budgets.
Good at
- Instruction following
- Text generation
- Reasoning
Limits
- Large context window requires significant VRAM
- No vision or audio support in this variant
- Requires llama.cpp 0.8.0+ for Gemma 4 MTP
Best for
- Chatbots needing long context
- Coding assistants with large prompts
- Research prototyping on GPUs
Who it’s for. Developers and researchers who need an instruction‑tuned Gemma 4 model with a large context window and a permissive license. The model is suitable for deployment on GPUs with moderate VRAM and for use with llama.cpp.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ3_M | 0.2 GB | 1.8 GB | Fits | Get file → | |
| IQ4_NL | 0.3 GB | 1.8 GB | Fits | Get file → | |
| Q4_0 | 0.3 GB | 1.8 GB | Fits | Get file → | |
| Q4_0 | 0.3 GB | 1.9 GB | Fits | Get file → | |
| Q8_0 | 0.5 GB | 2.0 GB | Fits | Get file → | |
| F16 | 0.9 GB | 2.4 GB | Fits | Get file → |
For 8 GB VRAM, use the IQ3_M or IQ4_NL quantizations (≈0.23–0.25 GB file, ~1.8 GB memory at 8 k context). For 12 GB VRAM, Q4_0 or Q4_0-q4emb (0.25–0.32 GB file, ~1.8–1.9 GB memory). If you have 16 GB or more, Q8_0 (0.46 GB file, ~2 GB memory) or F16 (0.86 GB file, ~2.4 GB memory) provide higher precision.
What GPU do you need for Assistant?
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 F16 (0.9 GB file, about 2.4 GB in use).
Which Assistant quant should you choose?
Q4_0 (0.3 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is IQ3_M at 0.2 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about Assistant
How much VRAM does Assistant need?
About 1.8 GB at Q4_0 with an 8K context window. That is the 0.3 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 Assistant on an 8 GB GPU?
Yes, with F16: the 0.9 GB file needs about 2.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Assistant GGUF?
It depends on the quant: IQ3_M is 0.2 GB, IQ4_NL is 0.3 GB, Q4_0 is 0.3 GB, Q4_0 is 0.3 GB, Q8_0 is 0.5 GB, F16 is 0.9 GB.
What context length does Assistant support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.
Where can I download Assistant GGUF?
The files are on Hugging Face in the RachidAR/gemma-4-26B-A4B-it-qat-assistant-q4_0-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 Assistant use?
Hugging Face lists the license as mit. 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 this model on 8 GB VRAM?
Yes, using the IQ3_M or IQ4_NL quantizations which use about 1.8 GB of VRAM at 8 k context.
Does this model support image input?
No, the Assistant variant does not include vision or audio support.
What is the context length?
The model supports up to 131 072 tokens of context.