What happened
On 16 November 2026 OpenAI published “More ways to move care forward with ChatGPT for Healthcare”, describing three additions: computer-use agents for clinical and administrative workflows, an improved “trusted clinical search” experience, and more than 20 new healthcare plugins. A second OpenAI post on computer-use agents for healthcare teams could not be read, so this article relies only on the first page. Everything below is as OpenAI describes it.
Key details
- Computer use: agents read screens, click, type and move between applications, with access controls and audit logs. OpenAI names Abridge, Ambience, Cigna Healthcare and CVS Health among the customers and partners, with CVS using it for processing recommendations and moving documents.
- Clinical search: in a blind evaluation, physicians rated answers to real clinical questions, with more than 1,300 ratings. OpenAI says the new experience scored better on all eight dimensions, including factual accuracy, safety, applicability and source authority. The post gives no numeric scores or full method.
- Plugins: 21 are listed (for example Availity Extend, Elation Health, Redox, UpToDate Expert AI and PenguinAI Medical Coding) and are available “today”, including through ChartGPT for electronic health records. Plugins need to be enabled by an administrator.
- Availability: for ChatGPT for Healthcare customers, depending on workspace settings. ChatGPT for Clinicians is free for eligible, verified US clinicians. Computer use is also in the API, ChatGPT Work and Codex, with a sales contact for clinical workflows. OpenAI offers a business associate agreement for HIPAA workflows and says content is not used for training.
- Not stated: pricing, numeric evaluation results, and availability outside the US (only the US is mentioned for Clinicians).
Why it matters
The change is that agents can now operate the screens of existing healthcare software, which is how much of hospital and payer work is actually done, rather than only answer questions. The evaluation claims are OpenAI’s own and unscored, so organisations should run their own tests, and any use that touches patient care needs clinical and compliance review.
This article is a product description, not medical advice.
Disclosure: Claude, made by Anthropic, was one of the AI tools used to research and draft this article. Anthropic competes with OpenAI; all claims are OpenAI’s own.
Sources
- More ways to move care forward with ChatGPT for Healthcare Primary source , OpenAI, 16 November 2026