Meridian Labs is a 501(c)(3) non-profit building an open source platform for understanding, evaluating, and testing AI models and agents. Our tools enable rigorous model testing and in-depth analysis of agent behavior.
Framework for systematic evaluation of large language models. Built in collaboration with the UK AISI.
In-depth analysis of AI agent transcripts with rich visualization and high-performance parallel scanning.
Synchronous monitoring for control and oversight of AI agents during training, evaluation, and deployment.
Automated multi-turn agent for alignment auditing of language models. Built in collaboration with the UK AISI.