4-bit Open on Hugging Face
Parameters
12B
Weights (est.)
≈ 6.7 GB
Mac memory (est.)
16 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Details
- Base model
- google/gemma-4-12B (quantized)
- Vendor
- Google · Gemma
- Task
- Any to any
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 898 in the last 30 days · 16K all time · 14 likes
- Published
- 2026-06-03 · updated 2026-06-03 (5mo ago)
- Tags
- gemma4_unified
Other MLX builds of google/gemma-4-12B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| gemma-4-12B-4bit (this one) | 4-bit | ≈ 6.7 GB | 16 GB+ | 898 |
| gemma-4-12B-mxfp4 | MXFP4 | ≈ 6.7 GB | 16 GB+ | 92 |
| gemma-4-12B-nvfp4 | 4-bit | ≈ 6.7 GB | 16 GB+ | 83 |
| gemma-4-12B-5bit | 5-bit | ≈ 8.2 GB | 16 GB+ | 71 |
| gemma-4-12B-6bit | 6-bit | ≈ 9.7 GB | 16 GB+ | 123 |
| gemma-4-12B-8bit | 8-bit | ≈ 12.7 GB | 24 GB+ | 203 |
| gemma-4-12B-mxfp8 | MXFP8 | ≈ 12.7 GB | 24 GB+ | 83 |
| gemma-4-12B-bf16 | bf16 | ≈ 23.9 GB | 32 GB+ | 139 |