This is one of those rare moments where clinician input was truly centered in the design of a health tech product. OpenAI is committed to unlocking the potential of LLMs to improve health outcomes, and empowering clinicians is a vital step! ❤️ 🩺
Today we’re introducing two big steps for health at OpenAI:
- ChatGPT for Clinicians, a free version of ChatGPT designed for clinical work
- HealthBench Professional, a new benchmark to evaluate real clinician chat tasks
We’re excited about what this can unlock for care. ❤️
In “conversations that reflect real-world use of LLMs, GPT-5.4 correctly recommends immediate care in emergency cases more than 99% of the time”. This team is committed to evaluation integrity and improvement because they care about the real outcomes here ❤️ ⚕️
Empowering individuals to understand, participate in and advocate for their care is one of the most powerful ways LLM tools will improve the health of the world ❤️🩹
There is a lot to consider when someone asks a question about their health. Efforts to improve model performance in this domain are truly making a difference in the health of the world. #Healthbench
GPT-5 is our best model yet for health-related questions, empowering users to be informed about and advocate for their health.
The model now provides more precise and reliable responses, adapting to the user’s context, knowledge level, and geography.
Empowering individuals to understand, participate in and advocate for their care is one of the most powerful ways LLM tools will improve the health of the world ❤️🩹
Last October, my wife was diagnosed with three cancers in a single week.
Today, we told that story on the GPT-5 livestream with @sama.
Not because it’s easy to tell, but because someone facing that news today should know there’s a new kind of lifeline.
🧵
@Felipe_Millon@sama Wishing Carolina all the best ❤️🙏 I’m grateful for our team and the opportunity to contribute to making tough health situations a little easier to navigate.
@thekaransinghal This is what it’s all about - helping people in real-world health situations. Outcomes are simply better when individuals understand, participate in and advocate for their care!
@fidjissimo @rahularoradfs It’s exciting to see what’s possible when you pair a caring clinician with tools like AI Consult. The potential to support clinicians and improve health outcomes is immense!
@gdb @rahularoradfs @PendaHealth It’s exciting to see what’s possible when you pair a caring clinician with tools like AI Consult. The potential to support clinicians and improve health outcomes is immense!
@kevinnbass Spaces exist where these fields come together - and there is so much good that can be done here for improving health!!
https://t.co/LWdNN4BwNb
The potential to support frontline clinicians with AI - not to replace, but to uplift - is immense!
A caring clinician + the right AI tool can = better care in real-world clinical settings.
I’m so proud to be part of this work and grateful to this team for making it real!
@thekaransinghal @rahularoradfs @rldistler@doctorroko
@thekaransinghal@PendaHealth It’s exciting to see what’s possible when you pair compassionate care with tools like AI Consult. Thanks for making space for innovation that truly supports clinicians on the frontline of health!
The potential to support frontline clinicians with AI - not to replace, but to uplift - is immense!
A caring clinician + the right AI tool can = better care in real-world clinical settings.
I’m so proud to be part of this work and grateful to this team for making it real!
@thekaransinghal @rahularoradfs @rldistler@doctorroko
📣 Excited to share our real-world study of an LLM clinical copilot, a collab between @OpenAI and @PendaHealth.
Across 39,849 live patient visits, clinicians with AI had a 16% relative reduction in diagnostic errors and a 13% reduction in treatment errors vs. those without. 🧵
I’m incredibly proud to lead the physician team contributing to OpenAI’s #Healthbench and to have helped shape the framework for its design. Evaluating how AI performs on real health challenges is essential!
📣 Proud to share HealthBench, an open-source benchmark from our Health AI team at OpenAI, measuring LLM performance and safety across 5000 realistic health conversations. 🧵
Unlike previous narrow benchmarks, HealthBench enables meaningful open-ended evaluation through 48,562 unique physician-written rubric criteria spanning several health contexts (e.g., emergencies, global health) and behavioral dimensions (e.g., accuracy, instruction following, communication).
Blog, paper, code: https://t.co/NsSPeIoHZy
I look forward to listening every week, have been a loyal listener for 2 years. This one was hard to finish. @DavidSacks you have changed my perspective on many issues and I value your intelligence and communication style usually, but the constant interruptions and speaking over your guest was frankly unenjoyable and the inability to admit that Trump has any shortcomings makes you less credible - he has some obvious flaws, it’s ok to admit them. @mcuban has some strange takes and theories, true, but he also was able to voice both pros and cons of his position and candidate - this is what moderates and undecided voters want and connect to. He is also doing some amazing work that I wanted to hear more about, but the conversation was so overtaken by politics. 🫤