Privacy-first local AI model comparison platform with blind evaluation, per-model hyperparameters, and multi-configuration testing. Compare 2-6 models side-by-side through Ollama with zero cloud dependencies.