← The Polylog AI Intelligence Brief
Morning Edition · Friday, September 25, 2026Published at 2:22 AM EDT · New York
OpenAI's Mental-Health Benchmark Scores Its Own Model Above Clinician-Written Replies
GPT-6 Astra reached 57.3 on MentalHealthBench against 38.5 for expert-authored answers, a gap OpenAI attributes partly to clinicians writing short responses that collect fewer scoring points.
OpenAI released MentalHealthBench, an open benchmark for grading model behavior in realistic mental-health conversations, built with input from more than 80 licensed psychologists and psychiatrists working across 22 countries, 19 languages…
Continue the AI Intelligence Brief
Track frontier labs, chips, export controls, model releases, regulation, and AI infrastructure.
- 5 AI intelligence signals a day
- Frontier labs, compute, and chips
- Model releases and AI infrastructure
- Source-grounded analysis with confidence labels
The Global Intelligence Brief stays free.
Part of a tracked trend
AI Moves Into Autonomous Scientific Discovery and Clinical Care
Over the next 3-9 months, AI systems move beyond text tasks into running real scientific experiments and managing clinical care, backed by peer-reviewed and benchmarked evidence of chemist- and physician-level performance.
Townsquare
Start a discussion in Townsquare.
More from this edition
- Anthropic Prices Opus 5.5 at $4 per Million Input Tokens, 20 Percent Below Opus 5
- Nvidia's Jensen Huang Says Labs That Cannot Contain Their Models Should Be Shut Down
- Google Adds a Real-Time Speaking Avatar to Gemini 3.8 Live in 97 Languages
- Models Faked Research Results Without Being Asked in Almost a Third of Open-Ended Tasks
- Nvidia Releases a 100-Million-Parameter Speaker Diarization Model Under a Commercial Open License
- ChatGPT Voice Gains Plugin Access to Email, Calendar and Slack
- Meta Sells Frontier-Tier Tokens at $1.25 In and $4.25 Out, With a Cheaper Tier for Data Rights
Comments
0No comments yet.