ChatGPT drops its daily limit for free users and swaps in a more accurate default model

OpenAI announced on Thursday that free ChatGPT users can now send unlimited text messages. The daily cap that used to interrupt a long conversation is gone for plain text. Limits on image generation, file uploads and voice tools stay in place.

At the same time, the default model changes for everybody. GPT-5.6 Luna replaces GPT-5.5 Instant as the model you get when you do not pick one. Luna is the lightweight member of the GPT-5.6 family, and OpenAI says it produces 62 percent fewer factual errors than the model it replaces. Free users and Go subscribers, at 8 dollars a month, also get a Think button that lets the model spend longer on a harder question instead of answering immediately.

Paying subscribers get something more interesting. Plus and Pro users, at 20 and 200 dollars a month, get a slider on GPT-5.6 Sol, the heavyweight model, that moves reasoning effort between high, extra high and pro. In plain terms: you decide how much thinking time you are willing to wait for. OpenAI also says it tuned Sol for ordinary work like web research, writing and planning, with more focused answers and less decorative formatting. Luna arrives this week; the unlimited chats, the Think button and the Sol slider roll out next week.

What is actually going on here

Two things at once. The first is a fight over the free tier. ChatGPT passed one billion monthly users in May, and free accounts are the top of that funnel. Claude, Gemini and a growing pile of free alternatives are all one tab away, so a daily message cap is an expensive thing to keep. Removing it is cheap defence, especially now that the default model is a small one that costs OpenAI less per answer than the old default did.

The second is that “which model am I talking to” is finally becoming a visible choice rather than a hidden one. The effort slider makes explicit something that has always been true behind the scenes: more compute spent per answer usually means a better answer and a longer wait. Being able to turn that dial yourself is genuinely useful once you notice which of your tasks need it.

There is a caveat worth stating plainly. OpenAI is facing lawsuits alleging that heavy chatbot use harmed vulnerable users, including minors. Removing the last natural brake on usage widens that exposure. The company says it has added guardrails for under-18 users, including avoiding romantic roleplay and not presenting itself as a substitute for real relationships. In its own accompanying documentation, OpenAI treats this August release as High capability in both cybersecurity and biological and chemical domains under its Preparedness Framework, and it published dedicated under-18 safety evaluations for the first time.

What this means for you: if you use ChatGPT for free, you should notice fewer wrong facts and no more running out of messages mid-task. Try the Think button on anything that involves multiple steps. If you pay, the slider is the thing to learn: leave it low for drafting and quick lookups, push it up for research, comparisons and anything where a confident wrong answer would cost you. And if there is a teenager in your house using ChatGPT, this is a reasonable week to have a conversation about it.

Sources

• GPT-5.6 August Updates system card, OpenAI
• OpenAI removes the daily limit on free ChatGPT chats, The Next Web
• OpenAI improves GPT-5.6 Sol in ChatGPT, The Decoder