LLM Playground

Multi-model experiments: debates, compression tests, and parallel evals across many models.