OpenAI gives free ChatGPT users unlimited text chats and a smarter default model
Free-tier users get GPT-5.6 Luna as their default model this week, plus unlimited text messaging next week. Paying subscribers get a more accurate GPT-5.6 Sol and a new control over how hard the AI thinks.
Key points
- OpenAI will remove text-message rate limits for free and Go tier ChatGPT users starting next week.
- Free and Go users are getting GPT-5.6 Luna as their default model this week.
- A new "Think" button, giving access to deeper reasoning for harder questions, arrives for free and Go users next week.
- GPT-5.6 Sol, the model for Plus and Pro subscribers, is being updated to make fewer factual errors, starting Thursday.
- Plus and Pro subscribers also get a new slider to control how much effort the AI puts into each answer, available Thursday.
OpenAI is expanding what its free users get from ChatGPT, the AI chatbot used by hundreds of millions of people worldwide. Two changes land this week and next, and together they close some of the gap between the free product and the paid tiers.
What is actually changing for free users?
Right now, free and Go tier users hit a ceiling after a certain number of text conversations. Next week that ceiling disappears for text. File uploads and images will still carry limits, but plain back-and-forth text chat will be unlimited.
This week, OpenAI is also swapping out the model that free and Go users talk to by default. The new default is GPT-5.6 Luna, a large language model (the technology that generates text responses in chatbots like ChatGPT). Luna is a recently released model that OpenAI describes as a step up from what free users had before.
Next week brings one more addition for those tiers: a "Think" button. Tap it, and ChatGPT applies more careful, step-by-step reasoning to a tricky question, the kind of processing that used to be gated behind paid plans.
What do paying subscribers get?
Plus and Pro subscribers (the paid tiers that cost $20 and $200 a month respectively) use GPT-5.6 Sol, a more capable model than Luna. Starting Thursday, Sol gets a targeted accuracy upgrade.
OpenAI says the updated Sol is "designed to make fewer mistakes, especially when answers depend on dates, numbers, sources, rules, or assumptions." It will also give shorter, more direct answers and skip padding when extra detail does not help.
The second addition for subscribers is a slider that lets users dial in how much "thought" ChatGPT applies to a response. Crank it up for a complex research question; leave it low for a quick reply to an email draft. Both changes go live Thursday, as first reported by The Verge AI.
What does this mean for ordinary users?
If you use the free version of ChatGPT for everyday tasks, writing help, or quick questions, the main practical change is simple: you can keep going without hitting a wall mid-conversation. The new Think button gives you a way to ask for a more careful answer when you really need one, without paying for a subscription.
For subscribers, the factual accuracy improvement matters most when you ask ChatGPT questions where getting a number or a date wrong would cause real problems. Think tax rules, medication schedules, or contract deadlines. The update does not make the model infallible, so checking anything high-stakes against a primary source is still sensible.
Common questions
Will free users get the same quality as paid users after these changes?
Not quite. Free users get GPT-5.6 Luna and a Think button; paying subscribers get the more capable GPT-5.6 Sol with stronger accuracy and the new reasoning slider. The gap narrows but does not close.
Do I need to do anything to get these updates?
No action needed. OpenAI rolls model and feature changes out automatically, so the new defaults and buttons will appear in ChatGPT on their own when the dates arrive.
