ChatGPT Free Limits Removed: What Changed for Free and Go Users

Since August 2026, ChatGPT's Free and Go plans have had no message limit for plain text chats. OpenAI announced the change on August 6 and at the same time switched the default model on these plans to GPT-5.6 Luna. "Unlimited", however, covers only text chat: features such as file uploads, image generation, voice mode and deep research keep their own usage limits.

What changed in ChatGPT's free limits?

According to the announcement as reported by the tech press, the change breaks down like this:

• Before: According to PCWorld, Free and Go users were limited to between 10 and 40 text messages on the flagship model in windows of 3 to 5 hours.
• Now: Text chats on the Free and Go plans have become unlimited. Engadget's review dated August 12 confirms that the change is in effect on free accounts.
• Model: The default model for Free and Go users changed from GPT-5.5 Instant to GPT-5.6 Luna. According to Engadget, free accounts can use only this model; paid plans get access to the more powerful models in the family.
• Think button: Free and Go users got a "Think" button that tells the model to reason longer on more complex questions. According to Webrazzi's report, use of this button is subject to separate limits.

Which ChatGPT features are still limited on the free plan?

It is easy to assume everything is open once the message limit goes, but that is not the case. As reported by TechCrunch, PCWorld and Engadget, separate limits remain for features that need more computing power:

• File and image uploads (chats with attachments)
• Image generation
• Voice mode
• Deep Research and similar tools

Engadget also notes that on the free plan the context window (the amount of text the model can consider at once) and the memory features are narrower than on paid plans. In practice, this means that even though the number of chats is unlimited, details from the start of a very long conversation can get lost over time.

What if you still see a "limit reached" warning?

The end of the message limit does not mean every warning has disappeared. The situation usually falls into one of these cases:

1. You used a non-text feature: If you added a file or image to the chat or generated an image, the limit may have been reached for that feature. Plain text chat is counted separately from that limit.
2. The conversation got too long: In very long chats, the model can mix up earlier details. The most practical fix is to ask for a summary of the conversation and start a new chat with that summary.
3. You used thinking mode: The "Think" button has its own limit; you can continue in normal chat without using it.

How do you get better results from unlimited text chat?

Now that the number of messages is no longer the problem, the real issue is the quality of the answer. Keeping in mind that the free plan has a single lightweight model and a narrower context window, these habits make a difference:

• Give the context up front: In the first message, write who you are, the purpose of the task, the target reader and the format you want. When the model does not have to guess, you need fewer corrections.
• Split large jobs: Instead of asking for a long report in one message, first work on the draft headings, then on each section separately.
• Open a new chat when the topic changes: Running different projects in the same chat leads the model to mix up earlier context.
• Verify figures and sources: The model can produce fluent but wrong information. Check dates, regulations and statistics in particular against the original source before publishing.

Two things to know about the free plan: ads and data

According to Engadget's roundup, the Free and Go plans show ad slots marked as "sponsored" below answers. The expansion of the free plan should therefore be read as part of an ad-funded model.

The second point is data. According to the same source, free accounts are included in model training by default, and this setting can be turned off under Data Controls. If your employees use ChatGPT with personal accounts, that setting and the content they enter are a direct risk for the company. For companies operating in Turkey, passing content that contains personal data, such as a customer's name, phone number or contract text, to an AI tool raises the company's obligations as data controller under Law No. 6698 on the Protection of Personal Data (KVKK). The law requires the data controller to take appropriate technical and administrative measures to prevent unlawful processing of and access to personal data. If you operate elsewhere, check the data protection rules that apply to you before staff paste customer data into any AI tool.

What does this mean for businesses?

Employees will use it more, so it is time to write rules

The message limit was a brake that in practice stopped many employees from using ChatGPT heavily. That brake is gone. If your company has no written rule on AI use, we recommend clarifying at least the following: which tasks it can be used for, which data must never be entered, who checks the output, and whether a company account or a personal account is used.

Content volume will grow, and originality will gain value

The cost of producing text has dropped to almost zero for free users. That means more blog posts, product descriptions and social media copy written in similar patterns. The content that stands out will be content that cites its sources, rests on experience and carries the brand's own data. AI can be used to speed up drafts; verification and final editing should stay with an editor.

Visibility in AI answers matters more

According to Webrazzi, ChatGPT has passed 1 billion weekly active users. As free use expands, people more often ask an AI assistant about a product, service or company before they use a search engine. Tracking whether your brand is described in these answers with correct, current information has become a new line of work alongside classic search engine optimization. We describe what we do here on our AI search optimization page, and our article on AI visibility and citations covers the basics.

In short

• On Free and Go, plain text chat has no message limit; the default model is GPT-5.6 Luna.
• Files, images, voice, deep research and the "Think" button are subject to separate limits.
• The free plan shows ads, and chats are open to model training by default; the setting can be turned off.
• For companies, the real agenda is a clear rule on employee use and discipline with personal data.

Plan details and limits can be changed by OpenAI; we recommend checking the current situation on your own account's plan page before you rely on it.