Openai launches Healthbench, a collection of healthcares for healthcare models

Openai, Creator Chatbot artificial intelligence ChatgptHe has A large open source language form called healthbench The company said in a blog post on Monday that this allows to measure models for the Agency for the Health Care Industry.

The model was built in partnership with 262 doctors in 60 countries, with 5,000 realistic health talks. The goal of Healthbench is to discover whether artificial intelligence models provide the best possible responses for health inquiries for people. Each response is measured against the percentage of the percentage of the percentage written by the doctor, with each criterion likely to match the doctor’s ruling. The model is recorded by GPT-4.1.

Openai’s O3 Thinking Form The best performance, according to Healthbench, with 60 %, followed by Elon Musk’s Groc 54 % and Google’s twin 2.5 Pro at 52 %.

In an example of an Openai’s blog, it offers a scenario where a 70 -year -old neighbor is lying on the floor, breathing but does not respond. A person asks artificial intelligence what to do. Then he provides an answer form with steps about what to do, such as calling emergency services, checking breathing and locating. Then Healthbench records the response, explain what the model answered correctly and what can be improved. Then it gives a final degree, in this case, 77 %.

The model can deal with 49 languages, including dazzhar and Nepalese, and includes 26 medical specialties, such as neurological surgery and eye medicine.

Openai did not immediately respond to a request for comment.