GE

Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic

GemmaGGUF by mradermacherUpdated 2026-10-13
Download on Hugging Face
Parameters11.9B
Context256K
Layers48
Architecturegemma4
Smallest quant4.8 GB
Quick answer

Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic needs about 20.8 GB of memory at Q4_K_M (a 7.4 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 11 quants, from 4.8 GB (Q2_K) to 12.7 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 51.5 GB.

Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic – a 12B parameter LLM with coding, reasoning, and vision abilities

Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic is a 11.9B parameter model in the Gemma family. It supports coding, reasoning, and vision tasks and is available in multiple GGUF quantizations. The model is licensed under Apache‑2.0 and has a trained context of 262144 tokens.

Good at

  • Coding tasks
  • Reasoning problems
  • Vision queries

Limits

  • No vision support in all quantizations
  • Requires large VRAM for higher‑quality quants
  • No mixture of experts

Best for

  • Developers needing code generation
  • Researchers testing reasoning benchmarks
  • Users running vision‑enabled inference

Who it’s for. The model is suitable for developers and researchers who need a versatile LLM with coding, reasoning, and vision capabilities. It is ideal for those who can manage the memory requirements of the larger quantizations.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K4.8 GB 18.2 GB
Too big Get file →
Q3_K_S5.5 GB 18.9 GB
Too big Get file →
Q3_K_M6.1 GB 19.5 GB
Too big Get file →
Q3_K_L6.6 GB 20.0 GB
Too big Get file →
IQ4_XS6.7 GB 20.1 GB
Too big Get file →
Q4_K_S7.0 GB 20.4 GB
Too big Get file →
Q4_K_M7.4 GB 20.8 GB
Too big Get file →
Q5_K_S8.3 GB 21.7 GB
Too big Get file →
Q5_K_M8.5 GB 21.9 GB
Too big Get file →
Q6_K9.8 GB 23.2 GB
Too big Get file →
Q8_012.7 GB 26.1 GB
Too big Get file →

For 8 GB VRAM, use Q2_K (4.83 GB file, 18.2 GB memory at 8 k context). For 12 GB VRAM, Q4_K_S (7.02 GB file, 20.4 GB memory) is a good balance. For 24 GB VRAM, Q6_K (9.79 GB file, 23.2 GB memory) provides higher quality.

What GPU do you need for Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored 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 Q6_K (9.8 GB file, about 23.2 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (12.7 GB file, about 26.1 GB in use).

Which Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic quant should you choose?

Q4_K_M is the usual starting point for Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic: it balances file size and quality well. Here it is a 7.4 GB file.

The Q8_0 file is 12.7 GB, about 1.7 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 4.8 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

Architecturegemma4
Parameters11.9B
Layers48
Trained context262,144 tokens
KV cache (fp16)1536 KB per token

Frequently asked questions about Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic

How much VRAM does Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic need?

About 20.8 GB at Q4_K_M with an 8K context window. That is the 7.4 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 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 18.2 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic GGUF?

It depends on the quant: Q2_K is 4.8 GB, Q3_K_S is 5.5 GB, Q3_K_M is 6.1 GB, Q3_K_L is 6.6 GB, IQ4_XS is 6.7 GB, Q4_K_S is 7.0 GB and 5 more.

What context length does Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored 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 12.9 GB at 8K and 51.5 GB at 32K.

Where can I download Gemma 4 12B Coder Fable5 Composer2.5 v1 Uncensored Heretic GGUF?

The files are on Hugging Face in the mradermacher/gemma-4-12B-coder-fable5-composer2.5-v1-uncensored-heretic-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 12B Coder Fable5 Composer2.5 v1 Uncensored 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 this model on 8 GB VRAM?

Yes, the Q2_K quant requires 18.2 GB memory at 8 k context, which is beyond 8 GB VRAM, so it cannot be run on 8 GB VRAM.

Does this model support vision tasks?

Yes, the model has detected vision ability, but the provided GGUF files do not include vision support.

What license does this model use?

The model is licensed under Apache‑2.0.

Other GGUFs of gemma-4-12B-coder-fable5-composer2.5-v1-uncensored-heretic

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Gemma models

All Gemma →

Other medium models

All medium models →