Grug 12B needs about 21.0 GB of memory at Q4_K_M (a 7.7 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 26 quants, from 0.0 GB to 23.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 51.5 GB.
Grug 12B – a compact‑reasoning fine‑tune of Gemma‑4‑12B‑it
Grug 12B is a 12‑billion‑parameter Gemma‑4 model fine‑tuned for concise reasoning. It reduces reasoning‑token usage while preserving answer quality. The model is available in multiple GGUF quantizations for llama.cpp and related tools.
Good at
- Compact reasoning traces
- Coding assistance
- Vision tasks
Limits
- Large context length requires high VRAM
- No explicit vision support in base model
- License is other, not open source
Best for
- Short answer generation
- Code debugging
- Rapid reasoning queries
Who it’s for. Developers and researchers who need a 12B model that uses fewer tokens for reasoning. It is suitable for local deployment with sufficient VRAM and for use in tools that support GGUF formats.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 13.4 GB | Too big | Get file → | |
| IQ2_S | 4.7 GB | 18.1 GB | Too big | Get file → | |
| IQ2_M | 4.9 GB | 18.3 GB | Too big | Get file → | |
| Q2_K | 5.1 GB | 18.5 GB | Too big | Get file → | |
| IQ3_XXS | 5.1 GB | 18.5 GB | Too big | Get file → | |
| Q2_K_L | 5.3 GB | 18.7 GB | Too big | Get file → | |
| IQ3_XS | 5.5 GB | 18.9 GB | Too big | Get file → | |
| Q3_K_S | 5.7 GB | 19.1 GB | Too big | Get file → | |
| IQ3_M | 6.0 GB | 19.4 GB | Too big | Get file → | |
| Q3_K_M | 6.3 GB | 19.7 GB | Too big | Get file → | |
| Q3_K_L | 6.7 GB | 20.0 GB | Too big | Get file → | |
| IQ4_XS | 6.8 GB | 20.2 GB | Too big | Get file → | |
| Q3_K_XL | 6.9 GB | 20.3 GB | Too big | Get file → | |
| IQ4_NL | 7.1 GB | 20.5 GB | Too big | Get file → | |
| Q4_0 | 7.1 GB | 20.5 GB | Too big | Get file → | |
| Q4_K_S | 7.2 GB | 20.6 GB | Too big | Get file → | |
| Q4_K_M | 7.7 GB | 21.0 GB | Too big | Get file → | |
| Q4_1 | 7.8 GB | 21.1 GB | Too big | Get file → | |
| Q4_K_L | 7.9 GB | 21.3 GB | Too big | Get file → | |
| Q5_K_S | 8.4 GB | 21.8 GB | Too big | Get file → | |
| Q5_K_M | 8.8 GB | 22.2 GB | Too big | Get file → | |
| Q5_K_L | 9.0 GB | 22.4 GB | Too big | Get file → | |
| Q6_K | 10.2 GB | 23.6 GB | Too big | Get file → | |
| Q6_K_L | 10.5 GB | 23.9 GB | Too big | Get file → | |
| Q8_0 | 12.7 GB | 26.1 GB | Too big | Get file → | |
| BF16 | 23.8 GB | 37.2 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (≈21 GB RAM estimate) or Q3_K_M (≈19.4 GB). For 12 GB VRAM, Q4_K_L (≈21.3 GB) or Q5_K_S (≈21.8 GB) are suitable. The smallest quant, IQ2_S, needs about 18.1 GB RAM but offers the lowest file size.
What GPU do you need for Grug 12B?
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: the best quant that fits is the main file (0.0 GB file, about 13.4 GB in use).
- 24 GB VRAM: the best quant that fits is Q6_K_L (10.5 GB file, about 23.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (12.7 GB file, about 26.1 GB in use).
- 48 GB VRAM: the best quant that fits is BF16 (23.8 GB file, about 37.2 GB in use).
Which Grug 12B quant should you choose?
Q4_K_M is the usual starting point for Grug 12B: it balances file size and quality well. Here it is a 7.7 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 the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Grug 12B
How much VRAM does Grug 12B need?
About 21.0 GB at Q4_K_M with an 8K context window. That is the 7.7 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 Grug 12B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 13.4 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Grug 12B GGUF?
It depends on the quant: file is 0.0 GB, IQ2_S is 4.7 GB, IQ2_M is 4.9 GB, Q2_K is 5.1 GB, IQ3_XXS is 5.1 GB, Q2_K_L is 5.3 GB and 20 more.
What context length does Grug 12B 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 Grug 12B GGUF?
The files are on Hugging Face in the bartowski/kai-os_Grug-12B-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 Grug 12B use?
Hugging Face lists the license as other. 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 Grug 12B on 8 GB VRAM?
The smallest quant IQ2_S requires about 18.1 GB RAM at 8k context, so 8 GB VRAM is insufficient for full context. A smaller context or lower‑bit quant may be needed.
Does Grug 12B support vision?
The base model includes vision capabilities, but the fine‑tuned version focuses on reasoning and coding; vision support is not emphasized.
What license applies to Grug 12B?
The model is released under an 'other' license, not a standard open‑source license.