The Free Version of ChatGPT Just Got a Real Upgrade
OpenAI is upgrading free ChatGPT with a smarter default model, unlimited text chats, and a Think button. What it means for small teams and what to try.
If your team runs on the free version of ChatGPT, you have probably treated it as the lite option: fine for a quick question, not something you would trust with real work. As of this week, that assumption is out of date.
On August 6, OpenAI announced an update that quietly changes what the free tier can do. Most of the coverage focused on the paid models. The part that matters for a small business or a nonprofit is what free users are getting.
What actually changed for free users
Three things.
A smarter default model. Free users are moving to a new default called GPT-5.6 Luna. In OpenAI's own internal testing on questions that turn on facts, dates, and numbers, Luna produced about 62% fewer answers with a factual error than the model free users had before. For anyone who has been burned by a confident, wrong AI answer, that is the number that counts.
Unlimited text chats. Free users have lived with usage caps, the point in the afternoon where ChatGPT tells you to wait or come back tomorrow. OpenAI is removing that limit for text conversations.
A "Think" button. For harder questions, free users get a button that tells the model to slow down and reason more carefully before answering, the kind of deeper thinking that used to sit behind the paywall.
The honest caveats
Two things worth knowing before you cancel anything.
First, the rollout is staged. The smarter default model lands this week. Unlimited chats and the Think button follow next week. So if you check today and it looks the same, give it a few days.
Second, "unlimited" means text chats. File uploads, image generation, and some other tools still have limits on the free plan. If your work leans heavily on uploading documents or generating images, the paid tiers still buy you something. For plain conversation, drafting, and thinking out loud, the free tier just got a lot more capable.
Why this matters for a small team
Most small organizations are not short on AI tools. They are short on time and budget. Until now, the free tier forced an awkward choice: ration your questions, pay for seats you were not sure you needed, or quietly decide AI was not worth the hassle for real work.
That calculation just shifted. A more reliable default plus no chat limit means the free version can handle the everyday work most 5-to-50-person teams actually have: drafting the donor email, untangling a confusing invoice, outlining the proposal, turning messy notes into a clean summary. Not every job, but far more than before. (If getting your team to actually use it is the real hurdle, that is a separate problem worth solving on its own.)
What to try this week
Do not overhaul anything. Run one experiment.
Pick a task you assumed needed the paid version or a person from start to finish. A first draft of a grant section. A spreadsheet whose logic you want explained in plain English. A plan you keep putting off. Run it through free ChatGPT and judge the result honestly against what you expected.
When the Think button arrives next week, save it for the harder questions, the ones where a fast, shallow answer would cost you. That is exactly what it is built for.
If the free tier holds up on a real task, you have just found capacity you were paying for or doing by hand. If it does not, you learned something specific and cheap, which beats another month of assuming.
What is one task you have been paying for AI to do that you would test on the free tier now?
Source: OpenAI: Improving GPT-5.6 Sol in ChatGPT and expanding access to GPT-5.6 Luna for free users (August 6, 2026).