GE

Gemma 4 26B A4B It Heretic

GemmaGGUF by nohurryUpdated 2026-04-04
Download on Hugging Face
Parameters25.2B
Context256K
Layers30
Architecturegemma4
Smallest quant12.7 GB
Quick answer

Gemma 4 26B A4B It Heretic needs about 25.4 GB of memory at Q4_K_M (a 16.9 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 13 quants, from 12.7 GB (Q3_K_S) to 100.9 GB (F32). Longer context needs more: at 32K tokens the KV cache alone adds about 32.2 GB.

Gemma 4 26B A4B It Heretic – a multimodal, Mixture‑of‑Experts LLM with 256K‑token context

Gemma 4 26B A4B It Heretic is a 26 billion‑parameter model that supports text and image input. It uses a Mixture‑of‑Experts architecture with 4 billion active parameters. The model is released under Apache 2.0 and is available in several GGUF quantizations.

Good at

  • Multimodal text‑image generation
  • Large‑context reasoning up to 256K tokens
  • Coding and function‑calling support

Limits

  • Requires 8 k context for memory estimates
  • No native audio support in this size
  • Large VRAM needed for higher‑quality quants

Best for

  • Developers building local multimodal assistants
  • Researchers testing 256K‑token context models
  • Edge deployments needing a 26B MoE model

Who it’s for. This model is suited for users who need a powerful, multimodal LLM that can handle long documents and code. It is ideal for developers and researchers who can allocate sufficient GPU memory and are comfortable with GGUF and llama.cpp workflows.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q3_K_S12.7 GB 21.3 GB
Too big Get file →
Q3_K_M13.9 GB 22.4 GB
Too big Get file →
IQ4_XS15.0 GB 23.5 GB
Too big Get file →
IQ4_NL15.4 GB 24.0 GB
Too big Get file →
Q4_K_S15.9 GB 24.5 GB
Too big Get file →
Q4_K_M16.9 GB 25.4 GB
Too big Get file →
Q5_K_S18.8 GB 27.3 GB
Too big Get file →
Q5_K_M21.2 GB 29.7 GB
Too big Get file →
Q6_K23.2 GB 31.7 GB
Too big Get file →
Q8_026.9 GB 35.4 GB
Too big Get file →
BF1650.5 GB 59.1 GB
Too big Get file →
F1650.5 GB 59.1 GB
Too big Get file →
F32100.9 GB 109.5 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_S (≈24.5 GB memory at 8 k context) or Q3_K_S (≈21.3 GB). For 12 GB VRAM, Q4_K_M (≈25.4 GB) or Q3_K_M (≈22.4 GB) are suitable. Higher‑quality quants such as Q5_K_M or Q6_K require 25–32 GB VRAM and are recommended only on high‑end GPUs.

What GPU do you need for Gemma 4 26B A4B It Heretic?

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_NL (15.4 GB file, about 24.0 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (23.2 GB file, about 31.7 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 A4B It Heretic quant should you choose?

Q4_K_M is the usual starting point for Gemma 4 26B A4B It Heretic: it balances file size and quality well. Here it is a 16.9 GB file.

The Q8_0 file is 26.9 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.

If memory is tight, the smallest file is Q3_K_S at 12.7 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

Architecturegemma4
Parameters25.2B (approximate, mixture of experts)
Active parameters4B
Layers30
Trained context262,144 tokens
KV cache (fp16)960 KB per token

Source

Quantized bynohurry
FamilyGemma
Licenseapache-2.0
Downloads (HF)4,129
Last updated2026-04-04

Frequently asked questions about Gemma 4 26B A4B It Heretic

How much VRAM does Gemma 4 26B A4B It Heretic need?

About 25.4 GB at Q4_K_M with an 8K context window. That is the 16.9 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 Gemma 4 26B A4B It Heretic on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 21.3 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Gemma 4 26B A4B It Heretic GGUF?

It depends on the quant: Q3_K_S is 12.7 GB, Q3_K_M is 13.9 GB, IQ4_XS is 15.0 GB, IQ4_NL is 15.4 GB, Q4_K_S is 15.9 GB, Q4_K_M is 16.9 GB and 7 more.

What context length does Gemma 4 26B A4B It Heretic 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 A4B It Heretic GGUF?

The files are on Hugging Face in the nohurry/gemma-4-26B-A4B-it-heretic-GUFF 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 A4B It Heretic 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 Gemma 4 26B It Heretic on 8 GB VRAM?

With Q4_K_S or Q3_K_S quants you can run the model, but memory usage will be close to the VRAM limit. Use 8 k context for best performance.

Does this model support vision tasks?

Yes, it can process image input but does not include native audio support in this size.

What license applies to this model?

The model is released under the Apache 2.0 license.

Other GGUFs of gemma-4-26B-A4B-it-heretic

Similar models

Browse all →

More from nohurry

All by nohurry →

More Gemma models

All Gemma →

Other medium models

All medium models →