Run google/gemma-4-12B-it-qat-q4_0-gguf locally
google/gemma-4-12B-it-qat-q4_0-gguf is a large instruction-tuned chat model with 11.91 billion parameters, built on the gemma4 architecture. It is released under the apache-2.0 license and has been downloaded 542,572 times.
To run google/gemma-4-12B-it-qat-q4_0-gguf locally at a 4,096-token context, its quantized versions need between 9.12 GB (Q4_0, lowest quality) and 9.12 GB (Q4_0, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is Q4_0, needing about 9.12 GB. That means google/gemma-4-12B-it-qat-q4_0-gguf fits entirely in the VRAM of a 10 GB GPU or larger, running fully on the GPU.
All quantizations
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| Q4_0 | 4.8 | Good | 6.66 GB | 1.67 GB | 9.12 GB | 7.5 t/s | Offload |
KV cache estimated (architecture unavailable). Speed is a rough estimate bounded by memory bandwidth.
Frequently asked questions
How much VRAM do you need to run google/gemma-4-12B-it-qat-q4_0-gguf?
You need about 9.12 GB of VRAM to run google/gemma-4-12B-it-qat-q4_0-gguf entirely on the GPU using the Q4_0 quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run google/gemma-4-12B-it-qat-q4_0-gguf on an 8 GB GPU?
Partially. google/gemma-4-12B-it-qat-q4_0-gguf only fits on an 8 GB GPU by offloading part of it to system RAM (with Q4_0), which runs but is slower.
Can I run google/gemma-4-12B-it-qat-q4_0-gguf on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run google/gemma-4-12B-it-qat-q4_0-gguf fully on the GPU using Q4_0 (about 9.12 GB).
Can I run google/gemma-4-12B-it-qat-q4_0-gguf on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run google/gemma-4-12B-it-qat-q4_0-gguf fully on the GPU using Q4_0 (about 9.12 GB).
What is the best quantization for google/gemma-4-12B-it-qat-q4_0-gguf?
If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.