Google Gemma4 12B needs about 20.5 GB of memory at Q4_K_M (a 7.1 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 3 quants, from 7.1 GB (Q4_K_M) to 23.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 51.5 GB.
Gemma4 12B Unified – multimodal LLM with text, image, audio, and video support
Gemma4 12B Unified is a 11.9 billion‑parameter model that handles text, image, audio, and video inputs. It offers a 262144‑token context window and multilingual support in over 140 languages. The model is released under Apache 2.0 and is available in several GGUF quantizations.
Good at
- Multimodal reasoning
- Tool calling with grammar‑constrained output
- Large context handling
Limits
- Vision encoder fails to load in current llama.cpp
- Requires recent llama.cpp for audio and vision
- No separate vision encoder
Best for
- Chat and instruction following
- Multimodal content generation
- Tool‑enabled workflows
Who it’s for. Developers and researchers who need a local, multimodal LLM with strong reasoning and tool‑calling capabilities. The model is suitable for laptop‑class hardware and can be run with the provided GGUF quantizations.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 7.1 GB | 20.5 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 (7.12 GB file, ~20.5 GB VRAM at 8 k context). For 12 GB VRAM, Q8_0 (12.67 GB file, ~26.1 GB VRAM). BF16 requires ~23.83 GB file and ~37.2 GB VRAM, suitable only for high‑end servers.
What GPU do you need for Google Gemma4 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: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (7.1 GB file, about 20.5 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 Google Gemma4 12B quant should you choose?
Q4_K_M is the usual starting point for Google Gemma4 12B: it balances file size and quality well. Here it is a 7.1 GB file.
The Q8_0 file is 12.7 GB, about 1.8 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.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Google Gemma4 12B
How much VRAM does Google Gemma4 12B need?
About 20.5 GB at Q4_K_M with an 8K context window. That is the 7.1 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 Google Gemma4 12B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 20.5 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Google Gemma4 12B GGUF?
It depends on the quant: Q4_K_M is 7.1 GB, Q8_0 is 12.7 GB, BF16 is 23.8 GB.
What context length does Google Gemma4 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 Google Gemma4 12B GGUF?
The files are on Hugging Face in the NobodyWho/Google_Gemma4-12B-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 Google Gemma4 12B 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 Gemma4 12B on 8 GB VRAM?
Yes, using the Q4_K_M quantization which requires about 20.5 GB VRAM at 8 k context, so it will not fit on 8 GB VRAM.
Does the model support vision?
The vision encoder is present but fails to load in the current llama.cpp build; a newer version is needed.
What license does the model use?
Apache 2.0, allowing free use and modification.