工具条件未记录 / max / 框架未记录 · Four independent responses per conversation; rubric score clipped to 0–100%; within-task mean then mean across 1,215 tasks.
MentalHealthBench · Overall; 1,215 synthetic conversations; four independent responses per task; clinician-authored rubrics · Mean task-clipped score · 工具条件未记录 / max / 框架未记录 · Four independent responses per conversation; rubric score clipped to 0–100%; within-task mean then mean across 1,215 tasks.
MentalHealthBench · Overall; 1,215 synthetic conversations; four independent responses per task; clinician-authored rubrics · Benchmark definition and scores
MentalHealthBench · Overall; 1,215 synthetic conversations; four independent responses per task; clinician-authored rubrics has benchmark disclosures or scores across 2 model releases.
Regular display scope. Rankings are grouped by metric and evaluation protocol. Recorded tool access, budgets, and frameworks are shown separately. Scores remain available for reference when conditions are unspecified.
Mean task-clipped score %
Tool access not recorded / max / Framework not recorded · Four independent responses per conversation; rubric score clipped to 0–100%; within-task mean then mean across 1,215 tasks.
MentalHealthBench · Overall; 1,215 synthetic conversations; four independent responses per task; clinician-authored rubrics · Mean task-clipped score · Tool access not recorded / max / Framework not recorded · Four independent responses per conversation; rubric score clipped to 0–100%; within-task mean then mean across 1,215 tasks.