HHH Alignment — leaderboard

Helpful, Honest, and Harmless alignment evaluation from BIG-bench/Anthropic-style criteria, testing preference for responses that balance usefulness, truthfulness, and safety.

Source: huggingface.co.

Interactive version: theaggregate.ai/benchmark?slug=hhh-alignment · How the rankings work · Data refreshed daily, snapshot 2026-07-22.