This research introduces CounselingBench, the first benchmark to systematically evaluate how well large language models align with core mental health counseling competencies.

• Tested 22 LLMs against professional standards using NCMHCE-based metrics
• Advanced models passed minimum aptitude thresholds but fell short of expert-level performance
• Models performed best at intake processes but struggled with assessment and diagnosis
• Identified specific gaps in therapeutic alliance and intervention competencies

Medical Impact: As healthcare faces a global shortage of mental health professionals, this research provides essential guidance on how LLMs might supplement clinical care safely and effectively, while highlighting critical limitations that must be addressed before deployment.

Do Large Language Models Align with Core Mental Health Counseling Competencies?