Ask once, compare many models.

LLM-MS runs multiple local models under one of two selection algorithms and shows the live competition as each model produces, scores, and converges.