NEWLLM & ChatbotsAnalysis
The Best Model on OpenAI's New Mental Health Benchmark Scores 57.3%
OpenAI's MentalHealthBench, built with 80-plus clinicians in 22 countries, tops out at 57.3% for GPT-6 Astra. Claude Opus 5.5 scored 52.4%.
Seung Jung·
6m#mental-health
NEW