Frontier health answers reach every free ChatGPT user
OpenAI has rolled GPT-5.5 Instant out to all free ChatGPT users and says the model now performs on its hardest health evaluations at a level comparable to its frontier Thinking models. The headline figures are a 71 per cent drop in factuality issues, better recognition of when a situation warrants urgent care, and clearer communication of uncertainty. The work was shaped by review from more than 260 physicians across 60 countries.
The scale of what this touches is worth sitting with. More than 230 million people each week ask ChatGPT about lab results, appointments and symptoms. Moving this capability below the paywall changes who gets reliable guidance, not just how good it is. If you have previously been cautious about trusting a chatbot on health questions, the physician-reviewed process behind this update is the relevant detail to weigh.
New spend controls and usage analytics land for ChatGPT Enterprise
OpenAI has switched on credit usage analytics and revamped spend controls inside the Global Admin Console for ChatGPT Enterprise. Admins can now see consumption broken down by user, product and model in a single dashboard, set a default workspace limit, configure limits per group, and create individual overrides for heavier users. Employees can view their own usage against budget and request additional credits with context attached.
For any business that has watched its AI bill climb without clear visibility, this is the sort of unglamorous but practical change that makes the platform genuinely easier to govern. If you administer ChatGPT for a team, the new controls are available in the console today.
OpenAI published applied research on using AI to help physicians diagnose rare genetic diseases in children, part of a wider push into scientific and clinical use cases. Research rather than a tool you can pick up today. OpenAI Newsroom, 18 Jun →
No product releases from Anthropic in the last 24 hours. However, access to Claude Fable 5 and Mythos 5 remains suspended under a US export-control directive, now entering its sixth day. At the Seoul office launch on 17 June, Anthropic executives said they were confident the models would return within days, but no restoration had been confirmed as of 18 June. If you rely on Fable 5 for long-horizon agentic work, treat its availability as still unresolved. TechTimes, 18 Jun →
Today's themes
The day's verified product movement came entirely from OpenAI, and both items pushed capability outward: frontier-grade health answers to every free user, and granular cost governance to enterprise admins. The direction is consistent: make the top of the capability stack available further down the pricing ladder, while giving the people who pay the most better tools to understand what they are spending. The contrast with Anthropic is the second theme worth sitting with, because access itself has become a variable: Fable 5 stayed offline under a government directive while OpenAI shipped, a reminder that frontier model availability is now partly a matter of policy. A sourcing note worth adding: two of the loudest stories circulating today, a Gemini 2.5 Flash default switch and a Fable 5 restoration, did not hold up against primary sources. It pays to check company channels before rebuilding a workflow around a headline.