ChatGPT free unlimited: what actually is, and what is not

On 6 August 2026, OpenAI published a four-line release note that removed ChatGPT's most disliked constraint: Free and Go users would get unlimited text chats. No more wall after a few dozen messages, no more countdown to a quota reset.

Three weeks on, it is live, and OpenAI's pricing page states it plainly on the Free card: unlimited written conversations with GPT-5.6 Luna. So the question is no longer whether it is true — it is. The question is what the word "unlimited" actually covers, because it carries three caveats, and one of them appears nowhere on the pricing page. It is buried in the ads FAQ.

Everything below comes from pages OpenAI publishes, all of them dated, and every claim points to its source. Where OpenAI gives no number — which is the case for almost every quota — I say so instead of repeating a figure that circulates elsewhere.

What changed on 6 August 2026

The 6 August release note runs to two sentences. The first says GPT-5.6 Luna becomes the default model for Free and Go users. The second says that, starting the following week, those same users would also get "unlimited text chats" and a new Think button for harder questions, subject to abuse guardrails.

And OpenAI adds the restriction immediately, in the same note: "Limits will still apply for file uploads, images and other tools." That sentence is the heart of the matter, and it is the one that travelled least.

Here is the full sequence, as it appears in OpenAI's release notes and help pages.

It is worth pausing on what this represents. For three years, ChatGPT's free version was defined by its ceiling: a message count, then a wait message. That ceiling is what pushed people to subscribe, and it is what fed the whole ecosystem of "free unlimited ChatGPT, no sign-up" sites that still occupy Google's first page. On text, that ceiling no longer exists — and those sites no longer have anything to sell on that particular point.

"Unlimited" means text, on a single model

First caveat, and it is structural: the unlimited applies to one model. Free and Go run on GPT-5.6 Luna, which OpenAI describes as "the fastest and cheapest model in the GPT-5.6 family". That is not a coincidence — an unlimited model has to be a cheap model to serve.

What the free tier cannot reach

The GPT-5.6 help page is explicit: "Free and Go users get GPT-5.6 Luna and do not have access to GPT-5.6 Sol." Sol is the model built for complex coding, knowledge work and research, cybersecurity and design tasks. It powers the reasoning slider on paid plans.

Worth noting in passing, because the question comes up constantly: logged-out users do not get GPT-5.6 Sol either. Using ChatGPT without an account does not hand you a degraded version of the best model; it hands you the same model as the free tier.

The context window, the limit nobody sees

Second caveat, and this one is invisible in the interface. The comparison table on the pricing page, Free column, shows a 27K context window for instant mode, with a maximum user input estimated at around 12 pages of text.

In practice: you can have an unlimited number of conversations, but each one has a short memory. In a long thread, ChatGPT loses the beginning. Paste a thirty-page document and it will not see all of it. OpenAI also notes that the share genuinely available for your input is smaller than the total window, because system instructions, memories and internal processing have to fit in there too.

The same column describes response speed as "limited availability and throughput". Unlimited in volume, then — but not first in the queue.

What is still capped — and what OpenAI will not quantify

This is the part most coverage skipped. Here is the exact sentence from the free tier FAQ: "Free users have unlimited everyday text chats, subject to abuse-prevention safeguards. File uploads, image generation, voice, data analysis, and other tools have separate usage limits. ChatGPT will notify you when you reach an applicable limit."

The word that matters is "separate". These limits are not a slice of the main limit — they are their own counters, living their own lives.

The three tools that fall together

One detail in the FAQ deserves a second reading. OpenAI lists three tools that carry limits separate from the main chat limit — data analysis, file and image uploads, and image creation — then adds: "If you reach a rate limit for any of these tools, you will need to wait for a later time to use any of these tools."

In other words, those three share a common wall. Generating a batch of images in the morning can stop you having a spreadsheet analysed in the afternoon. That is exactly the kind of behaviour you discover at the worst possible moment, and it is written down in the documentation.

GPTs: usable, not creatable

The free tier lets you discover and use existing GPTs. Their usage "follows your current Free tier model limit": when you reach it, GPT access pauses until the limit resets, and ChatGPT shows the reset timing.

What surprises many people, though: GPT creation is no longer available on any personal account — Free, Go, Plus and Pro included. OpenAI now reserves it for Business, Enterprise and Edu workspaces. Existing GPTs remain usable, and editing one still depends on your plan and permissions.

The catch: unlimited is tied to advertising

This is the most interesting caveat, and it is not on the pricing page at all. You have to go and find it in the ads FAQ.

Context first: since 24 August 2026, ChatGPT Ads is open across 31 European markets. Ads may appear for users on the Free and Go plans. Plus, Pro, Business, Enterprise and Edu do not include ads, and OpenAI does not serve them to accounts identified as belonging to people under 18. I covered how the ad platform works in the article on the European ChatGPT Ads launch.

The FAQ then asks the question everyone asks: "Will there be a way to not see ads?" OpenAI's answer, word for word:

Yes. You can reduce or avoid ads in a few ways: upgrade to an ad-free subscription plan (Plus or Pro); choose a free, no-ads option with additional limits (fewer messages per day and no access to tools like images).

Read that second branch again. "Fewer messages per day": you are back to a daily ceiling. "No access to tools like images": that is not a limit any more, it is a removal.

The conclusion is simple, and nobody is spelling it out: unlimited text is the ad-supported free tier. This is not a gift from OpenAI, it is the free version's new business model. You no longer pay in rationed messages, you pay in attention.

Two useful notes before you decide. First, OpenAI adds that "availability may vary by region and over time": I could not verify in which countries the free no-ads option is actually offered, and OpenAI publishes no list. Second, Temporary Chats show no ads on any plan — a way to silence a single ad without touching your settings.

Think on the free tier, and what it does not do

August's other change is the Think button, now available to Free and Go users for harder questions. On mobile you open the + menu and select Think; on the web it landed in the 14 August release note. The free tier FAQ still described the web rollout as in progress at the time of writing, so if you cannot see it, that is probably why.

Here is the nuance worth knowing, straight from the GPT-5.6 help page: "Think uses GPT-5.6 Luna, not GPT-5.6 Sol." So this is not the paid plans' reasoning mode quietly unlocked for free. It is the same fast model, asked to take longer.

What the free tier does not get, on the other hand:

• the reasoning slider and its Medium, High, Extra high and Pro levels, all built on GPT-5.6 Sol;
• automatic reasoning, which lets ChatGPT go deeper on its own for a complex query and is toggled under Settings, then General, then Higher intelligence;
• the fallback to another reasoning model when a limit is reached, since there is no other reasoning model available.

Free, Go, Plus: what a paid plan actually buys now

With text unlimited at the bottom of the range, the subscription question looks different. Here are the official prices as OpenAI displays them on its European pricing page on the day this article was published. Prices are shown per country and per currency, so the figures you see may differ from these.

Two things follow from that table.

First: Go no longer buys unlimited text, because the free tier already has it. Go buys volume on the tools, and nothing else. If you were considering Go so you would "stop getting cut off mid-conversation", that argument died in early August. If you were considering it because you max out image generation or uploads, it still stands.

Second: Go still carries ads. It is the only paid plan in that position. Paying for Go does not remove advertising; you have to go up to Plus for that. Worth knowing before you click.

One documented tip for anyone on the fence: if you hit a limit on the free tier, upgrading to Plus, Pro or Business resets your usage rates immediately. You do not wait out the current window.

What this changes for you in practice

If you use ChatGPT to write and think

The free tier has become good enough, and that is new. Drafting, rewriting, translating, outlining, exploring a topic, asking question after question: none of it burns a quota any more. The one constraint left is the 27K context window, which forces you to split very long documents and to start a fresh conversation when a thread drags on.

If you work with files and images

Nothing has changed for you, and that is the point. The counters that used to block you are exactly the ones that blocked you in July. Worse, they are interlocked: maxing out one of the three tools closes the other two. If that is your main use, the 6 August announcement does not concern you, and the real decision sits between Go and Plus.

If ads bother you

You have three doors, and they are not equivalent. Upgrading to Plus or Pro removes ads in exchange for a subscription. Using a Temporary Chat removes them for that conversation only, changing nothing about your account. And the free no-ads option removes them in exchange for a daily message ceiling and the loss of image tools — which is, for all practical purposes, a return to July.

If you were shopping for a free alternative

The comparison has moved. While ChatGPT's free tier was capped, the first criterion was the message count. Now that text is unlimited, that criterion disappears and the real differences shift elsewhere: the quality of the model offered for free, the size of the context window, and what each service does with your conversations. That is the angle to re-read Claude's free plan and the free AI tools comparison with.

FAQ - Frequently asked questions

The questions that have come up most since the 6 August announcement, answered briefly and sourced.

The bottom line

The announcement is real and it matters: the message cap that has defined ChatGPT's free version since launch is gone for text. That is not a detail, and it makes the free tier genuinely usable for anything involving writing, translating and thinking out loud.

But "unlimited" here describes exactly one thing: text, on the lightest model in the range, inside a short context window, on a plan funded by advertising. Everything expensive to serve — images, files, voice, data analysis, deep research — stays capped, without OpenAI publishing a single number. And if you refuse the ads, you get a daily ceiling back and lose the image tools.

So the question to ask is no longer "will I get cut off?" but "what will I get cut off on?". For many people the answer will be: nothing. For anyone working with files, it is exactly the same answer as before 6 August.

Want to follow this story? Subscribe to the EverydayAITech newsletter for weekly analysis in plain language.

---

About the author: Flavien Hue has been testing and analysing artificial intelligence tools since 2023. His mission: making AI accessible through practical, honest guides, without unnecessary jargon.