PromptLens benchmarks

Evaluation evidence, not universal leaderboards

Worked LLM evaluation studies with visible datasets, scoring conditions, failures, and release decisions. Every study states whether its data is demo-safe, synthetic, or production-derived.