Clinical decision-support platform OpenEvidence launched a new family of AI models for physicians on September 3, 2026, deepening its push toward what founder Daniel Nadler calls "medical super-intelligence" for the more than 915,000 medical-license-verified U.S. clinicians already on its platform.
Specialty models under one platform
Nadler has described the strategy as layering AI specialty models — oncology, neurology and others — beneath a single clinician-facing interface, so front-line users get evidence-grounded answers tuned for their subspecialty. The new family expands on an increasingly crowded clinical-AI landscape that also includes ChatGPT medical modes, Microsoft's MAI-DxO diagnostic orchestrator, iatroX and Medwise.
Grading the evidence, and now the calls
The launch follows OpenEvidence's July rollout of EvidenceGrade, a real-time source-quality grader that tags answers by publication strength, and an AI-integrated dialer that reads a caller's clinical context aloud to physicians. Earlier this summer the company disclosed a NewYork-Presbyterian collaboration and hands-free workflow features for 860,000 clinicians.
Riding the regulatory wave
OpenEvidence's launch lands the same week as the FDA's TEMPO pilot for generative-AI medical devices, which provisionally allows some developers to launch products while regulators refine the framework. Combined with Nature Medicine research this year showing that general-purpose LLMs are increasingly competitive with specialized clinical AI, the announcements underscore how fast the front line of medicine is being reshaped by foundation-model tooling.
Reporting based on coverage from STAT, Fierce Healthcare, PYMNTS and HIT Consultant.