Gemma 4 26B MoE needs about 27.7 GB of memory at Q5_K_M (a 19.1 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 5 quants, from 12.4 GB (IQ3_M) to 26.9 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 32.2 GB.
Gemma 4 26B Mixture‑of‑Experts with custom importance matrix
Gemma 4 26B MoE is a 25.2B‑parameter model with 128 experts. It supports a 262144‑token context window and uses a custom importance matrix to maintain routing stability at lower bit‑rates. The model is released under apache‑2.0.
Good at
- Long‑context reasoning
- High‑parameter efficiency
- Expert routing precision
Limits
- No vision support
- Requires 24GB VRAM for full precision
- Large memory footprint for higher quants
Best for
- Enterprise chatbots
- Research prototypes
- Large‑scale text generation
Who it’s for. Developers and researchers needing a large‑parameter LLM with efficient expert routing. The model is suitable for workloads that can accommodate its memory demands and benefit from extended context windows.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ3_M | 12.4 GB | 20.9 GB | Too big | Get file → | |
| IQ4_XS | 13.9 GB | 22.5 GB | Too big | Get file → | |
| Q5_K_M | 19.1 GB | 27.7 GB | Too big | Get file → | |
| Q6_K | 22.6 GB | 31.2 GB | Too big | Get file → | |
| Q8_0 | 26.9 GB | 35.4 GB | Too big | Get file → |
For 8‑GB VRAM use IQ3_M (12.39 GB file, 20.9 GB memory). For 12‑GB VRAM use IQ4_XS (13.92 GB file, 22.5 GB memory). For 24‑GB VRAM use Q8_0 (26.86 GB file, 35.4 GB memory). Estimates are approximate.
What GPU do you need for Gemma 4 26B MoE?
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: 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: the best quant that fits is IQ4_XS (13.9 GB file, about 22.5 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (22.6 GB file, about 31.2 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (26.9 GB file, about 35.4 GB in use).
Which Gemma 4 26B MoE quant should you choose?
Q5_K_M (19.1 GB) is the middle option in this repository. Closer to the original model than 4-bit, for a moderate increase in size. A good pick when you have memory to spare.
If memory is tight, the smallest file is IQ3_M at 12.4 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
Source
Frequently asked questions about Gemma 4 26B MoE
How much VRAM does Gemma 4 26B MoE need?
About 27.7 GB at Q5_K_M with an 8K context window. That is the 19.1 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 Gemma 4 26B MoE on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 20.9 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Gemma 4 26B MoE GGUF?
It depends on the quant: IQ3_M is 12.4 GB, IQ4_XS is 13.9 GB, Q5_K_M is 19.1 GB, Q6_K is 22.6 GB, Q8_0 is 26.9 GB.
What context length does Gemma 4 26B MoE support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 8.1 GB at 8K and 32.2 GB at 32K.
Where can I download Gemma 4 26B MoE GGUF?
The files are on Hugging Face in the CelesteImperia/Gemma-4-26B-MoE-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 Gemma 4 26B MoE 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 this model on 8 GB VRAM?
Yes, the IQ3_M quant uses 12.39 GB file size and requires about 20.9 GB memory at 8k context, which fits an 8 GB VRAM setup with swapping.
Does this model support vision?
No, the model has no vision capabilities.
What license applies?
The model is released under the apache‑2.0 license.