Llama Bench

A Generally Adopted Benchmark for Local AI Inference Speed

llama-bench provides tokens-per-second metrics for local inference. Having a standard benchmark makes hardware and model comparisons meaningful instead of