Back to registry
METALUnified VRAM
arnav080/gemma-4-12b-q4-k-m-m1-16gb-balanced
Gemma 4 12B optimized for Apple M1 16GB. Reduced KV cache and prompt processing overhead for faster TTFT.
Configuration Specifications
| Base Model | huggingface:unsloth/gemma-4-12b-it |
| Engine | llama.cpp |
| Quantization | Q4_K_M |
| Platform | metal |
| VRAM Required | Unified minimum |
Telemetry Benchmarks
No runs recorded yet
Be the first to benchmark this recipe! Run the CLI command from your terminal:
bloc run arnav080/gemma-4-12b-q4-k-m-m1-16gb-balanced
gemma-4-12b-q4-k-m-m1-16gb-balanced.yaml
Loading workspace editor...