230m use ChatGPT for health — is your next diagnosis an algorithm?
openai's gpt-5.5 instant model for healthcare now outperforms doctors on accuracy benchmarks, signaling a major shift in how patients manage their own care.
openai has upgraded chatgpt with gpt-5.5 instant, a healthcare-optimized model developed using feedback from 260 physicians across 60 countries who reviewed 700,000 responses. the model matches expensive thinking models on machine-based health tests like healthbench and professional, while scoring higher than physicians on accuracy in openai's benchmarks. incorrect health statements reportedly dropped by 71% over two months, with the service now handling roughly 230 million weekly health queries from users. this expands ai's role from general assistant to a critical patient education tool, challenging traditional medical information channels.