Compare hosted LLM endpoints using public benchmark results. Inspect scores, latency, sample counts, prompts, outputs, and methodology across providers and models.