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 Modelhuggingface:unsloth/gemma-4-12b-it
Enginellama.cpp
QuantizationQ4_K_M
Platformmetal
VRAM RequiredUnified 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...