MentalHealthBench: open benchmark with mental health experts to evaluate AI responses

thekaransinghal · x · 2026-09-24

Ali Malik's team announced MentalHealthBench, a new open benchmark developed with mental health experts to evaluate how AI models respond in realistic mental health conversations.

Unlike simple Q&A benchmarks, it focuses on real-world conversational scenarios, probing model response quality and safety in sensitive support contexts. The project is open-sourced on GitHub with an explainer thread.

Related event: OpenAI Open-Sources MentalHealthBench for Evaluating AI Mental Health Support(12 posts)→

Original post →

More from Research

Research channel →