Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 10,525 prompts · See how Parse measures this
Meridian Labs is a 501(c)(3) nonprofit building an open-source platform to understand, evaluate, and test AI models and agents. Its tools enable rigorous model testing and in-depth analysis of agent behavior, including framework-based evaluation of large language models and transcript analysis with visualization. Projects include Inspect AI for systematic LLM evaluation, Inspect Scout for in-depth transcript analysis with visualization, and Inspect Petri for automated multi-turn alignment auditing, developed in collaboration with the UK AISI.
Parse Score
