Editorial illustration for Free ChatGPT Users Get Worse Health Advice From Older AI Model
Free ChatGPT Users Get Worse Health Advice
OpenAI began rolling out its "Health in ChatGPT" feature this week to U.S. users 18 and older, capping off testing that started back in January. The pitch sounds useful enough: connect Apple Health, upload medical records, sync wellness apps, and ChatGPT will help you make sense of lab results or get ready for a doctor's appointment. OpenAI says none of that connected data gets used for model training or advertising.
The catch is which version of ChatGPT actually answers your questions. Free users get GPT-5.5 Instant, a model that scores lower on health-specific benchmarks. Paying subscribers get GPT-5.6 Sol, the newer flagship OpenAI built to handle exactly this kind of query. So the same question about a blood test result or a medication interaction could get a meaningfully different answer depending on whether you're on a free or paid plan.
More than 300 million people ask ChatGPT health questions every week, according to OpenAI's own figures. That scale is exactly why the tiering matters, and why the benchmark numbers OpenAI leans on to justify the split deserve a closer look.
Users on the free version of ChatGPT receive lower-quality health advice. OpenAI powers the feature with GPT-5.5 Instant, which scores lower on health benchmarks than the new flagship model, GPT-5.6 Sol, reserved for paying users.
Why this matters
For developers and founders building on top of ChatGPT, this is a preview of how OpenAI will monetize trust. Health data is about as sensitive as it gets, and now there's a documented quality gap between free and paid tiers on exactly that category of advice. That's a tiering decision with real consequences: someone using the free version to interpret lab results isn't told they're getting a weaker model than a paying subscriber asking the same question.
If you're building health or wellness tools on OpenAI's API, this is worth watching closely, because model access tiers may start mapping directly to safety-relevant outcomes rather than just speed or context length. For researchers, it raises a benchmarking question OpenAI hasn't fully answered: what exactly separates GPT-5.5 Instant from GPT-5.6 Sol on health-specific tasks, and how big is that gap in practice? With 300 million-plus people already asking ChatGPT health questions, the stakes for getting this disclosure right, or wrong, are not small.
Expect regulators and competitors to start asking the same questions we are.
Related Reading
Browse all OpenAI news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic OpenAI hires 630 ex-Meta staff as ChatGPT memory may turn data into ads
Business & Startups
Related Topic Sen. Warren urges Trump admin to expand AI credit after OpenAI letter
Business & Startups
From Archives Experts: Kimi K3's Gains Not From Costly, Slow Frontier Model API
From Archives Cisco’s Small AI Models Outperform Larger Rivals on Cost for Vulnerability Detection
From Archives Apple Sues Ex-Executive After 24 Years, Alleges He Took AI Secrets to OpenAI
Policy & Regulation
From Archives OpenAI's Georgia Data Center Project Secures 3.2-Gigawatt Power Deal
Business & Startups
Common Questions Answered
What is the 'Health in ChatGPT' feature and what data can it access?
The 'Health in ChatGPT' feature allows users to connect Apple Health, upload medical records, and sync wellness apps so ChatGPT can help interpret lab results and prepare for doctor's appointments. OpenAI has stated that none of this connected health data gets used for model training or advertising purposes.
Why do free ChatGPT users receive lower-quality health advice than paid subscribers?
Free users receive health advice powered by GPT-3.5 Instant, which scores lower on health benchmarks compared to the flagship model GPT-4 Sol that is reserved for paying subscribers. This creates a documented quality gap where free users are not informed they're receiving weaker health guidance than premium users asking identical questions.
Which AI models power the health feature for different ChatGPT tiers?
OpenAI powers the Health in ChatGPT feature with GPT-3.5 Instant for free users and the newer flagship model GPT-4 Sol for paid subscribers. The performance difference between these models on health-related benchmarks means paying users receive more reliable health information.
What are the real-world consequences of the quality gap between free and paid health advice tiers?
The quality gap has serious implications because free users interpreting lab results or seeking health guidance don't know they're receiving advice from a weaker model than paying subscribers. This tiering decision on sensitive health data represents how OpenAI is monetizing trust, with potentially significant health consequences for users unaware of the model quality difference.
Further Reading
- OpenAI Brings Improved Health Responses To Free ChatGPT - Search Engine Journal
- Half of AI health answers are wrong even though they sound convincing - The Conversation
- Health Advice From A.I. Chatbots Is Frequently Wrong, Study Shows - The New York Times
- ChatGPT Health fails to recognise medical emergencies, experts warn - The Guardian
- Study finds popular AI chatbots often give problematic health advice - News-Medical.net
