On this page

On November 16, 2026, OpenAI published two product updates expanding ChatGPT for Healthcare and ChatGPT for Clinicians with computer-use agents, upgraded trusted clinical search, and more than 20 new healthcare plugins. The announcements focus on governed healthcare workflows, human-in-the-loop oversight, and benchmark disclosures.

Computer-Use Agents, Upgraded Clinical Search & 20+ Healthcare Plugins

Healthcare Computer-Use Agents (Drafts for Human Review): Available across the OpenAI API, ChatGPT for Healthcare, and ChatGPT for Clinicians (via ChatGPT Work and Codex), computer-use agents carry out multi-step workflows by reading screens, clicking, typing, and moving between existing healthcare applications. In OpenAI's demonstrations using synthetic patient data, agents review charts and prepare draft MRI orders, clinical notes, unsent patient messages, and prior authorization requests explicitly for clinician or specialist review and approval.

Improved Trusted Clinical Search: OpenAI updated how ChatGPT for Healthcare and ChatGPT for Clinicians find, evaluate, and cite medical evidence. In a blind evaluation spanning over 1,300 physician reviews across multiple specialties and multi-turn conversations, physicians rated the updated experience higher across all eight evaluated dimensions (interactive chart values): Goal satisfaction (88.0% vs. 80.2%), Factual correctness (97.6% vs. 93.4%), Safety (95.5% vs. 89.4%), Actionability (92.4% vs. 86.8%), Calibration (95.6% vs. 92.8%), Authoritativeness (88.9% vs. 84.4%), Locality alignment (97.5% vs. 94.4%), and Temporal alignment (98.8% vs. 94.4%).

More Than 20 New Healthcare Plugins: All new plugins are available today in ChatGPT for Healthcare (subject to workspace admin enablement, including EHR workflow access via ChartGPT), with select plugins available to individual clinicians. Listed official plugin integrations include Aiva, Assort Health, Elation Health, Ferry Health, IMO Health, KMED, Luma Health, Notable, Oracle HDI, PenguinAI, Photon, Redox, Stedi, Subsalt, TriZetto, Turquoise Health, and UpToDate Expert AI.

Official Workflow Evaluations & Health System Deployments

EHR Inbox Evaluation (GPT-6 Astra): Evaluated at maximum reasoning effort on physician-shaped EHR inbox workflows, GPT-6 Astra was rated clinically safe (free of inaccuracies, omissions, or actions that could plausibly lead to patient harm) in 99% of 325 reviewed tasks, and rated very or extremely helpful in 93% of 646 physician assessments.

HealthAdminBench Administrative Benchmark (GPT-6 Astra & GPT-6 Luna): On HealthAdminBench, an independent benchmark from Stanford researchers covering prior authorizations, appeals, and medical equipment orders, OpenAI reports GPT-6 Astra scored 94.3% of available rubric points while GPT-6 Luna scored 91.8% at 8.5 cents per task (98% lower cost under OpenAI's benchmark measurement, evaluated at maximum reasoning effort).

Healthcare Partner Workflows: OpenAI highlighted deployments and prototypes with CVS Health (document routing, referrals, specialty consults, and an open-source Agent Kit built on the OpenAI Agents SDK), Cigna Healthcare (gathering case-management context across more than 10 systems), Ambience (pausing for clinician input and step-by-step audit logs), Assort (referral closure in Athena via Codex), and Abridge (planned inpatient care coordination).

Eligibility, BAA Support & Data Privacy

Governed Enterprise Workspaces vs. Free U.S. Clinician Access: ChatGPT for Healthcare provides role-based access controls and audit logs for health systems (available via sales or AWS Marketplace private offers), while ChatGPT for Clinicians is free for verified U.S. licensed clinicians (verified via NPI during sign-up per the product FAQ). Both products offer a Business Associate Agreement (BAA) to support HIPAA-compliant workflows when required security and retention settings are applied. OpenAI's product announcement states that content shared with ChatGPT for Healthcare is not used to train models, and the accompanying technical post states that organization data is not used to train OpenAI foundation models. The ChatGPT for Clinicians product page states that content shared with ChatGPT for Clinicians is not used to train OpenAI's models.