Benchmarking Local Models

Benchmark sovereign local LLM performance: tokens/second, memory usage, first-token latency, and quality metrics across Ollama models on Apple Silicon, NVIDIA, and AMD hardware.