OpenAI has decided to lift the limits on the number of text-based conversations for all ChatGPT users, including those on the free plan. The company announced the decision today, after ChatGPT achieved a milestone of 1 billion weekly users, which it also announced in the statement. Moreover, Open AI introduced a new default model, as well as several other changes to the free, paid, and professional plans.
Removing limits on text chats for free users is one of the most significant changes in the whole update, since limiting the number of conversations was one of the main ways Open AI pushed users to subscribe to the paid plan.
A New Default Model for Free and Go Users
The new default chat model will be GPT-5.6 Luna for Free and Go users. The new model will replace the GPT-5.5, which was previously the default.
In addition to the new default model, Free and Go users will see a new “Think” button that allows them to choose an enhanced model for more complex queries rather than using the default one. Notably, OpenAI stressed that although chats will be free and unlimited, other activities such as uploads, image creation, and voice-to-text conversation will have limits.
Thus, the clarification is vital for users who want to exploit the update to engage in unrestricted ChatGPT activities. The removal of restrictions applies only to chats. Other activities that require more effort from the AI servers will continue to have limitations set by the company.
Changes for Plus and Pro Subscribers
The change is not restricted to free users only, as OpenAI announced that its Plus and Pro subscription holders will also receive updates. In particular, they will get access to the improved model called GPT-5.6 Sol.
The company stated that the new model version is optimized to perform simple everyday tasks like answering questions, conducting research on the web, providing advice, planning, writing assistance, and decision-making. The changes made to GPT-5.6 Sol are supposed to make responses more concise and powerful than before.
Notably, this version of GPT-5.6 Sol is different from the one used in OpenAI’s Codex and Work products, which remains unchanged by this update. Overall, the update suggests that OpenAI rarely makes significant improvements to the same model version and typically launches new iterations instead.
Finally, Plus and Pro subscription holders will also get a new thinking slider that allows managing the amount of reasoning put into responses. The feature lets users adjust the slider based on how complex the query is, giving them more control over how much reasoning the model applies.
Fewer Factual Errors, According to OpenAI’s Internal Testing
OpenAI released results of an internal comparison between the new models and GPT-5.5-Instant. The latter is set to be replaced by newer alternatives in the upcoming versions of ChatGPT. According to OpenAI, factual errors were down 62% in GPT-5.6 Luna and down 68% in GPT-5.6 Sol, compared to GPT-5.5-Instant.
Both figures were reported as a result of internal research conducted by OpenAI. Therefore, users should be careful in evaluating the benefits demonstrated by the two models, as they only represent the performance relative to another system managed by the same parent company. It is common for firms to report such figures when releasing major updates to their products, but these results are generally not considered strong evidence of quality improvements.
Factual hallucinations are among the most prominent issues affecting various kinds of chatbots throughout their development. While modern AI systems can reason effectively in many abstract domains, they frequently produce false statements which appear to be reasonable from a logical perspective while containing demonstrably wrong information. In most cases, such flaws tend to emerge when the model attempts to provide an answer which requires specific factual knowledge. A significant reduction of such errors, even by a relatively small margin, would produce a major impact on the practical usefulness of any language model. This applies to ChatGPT in particular, as it is designed to serve as a helpful assistant which can be relied on for factual accuracy in its responses.
Rollout Timeline
Not all updates are released simultaneously. The new GPT-5.6 Sol version is now available to Plus and Pro users on the day of the update. Other modifications, such as the new default model and the Think button for Free and Go users, will be rolled out this week.
The availability of unlimited text chats and the Think button for free users is scheduled for the following week, so not all updates at once. It is a standard practice for major platforms to roll out big updates in stages in order to ensure that the platform will continue to function properly and will not disrupt the user experience.
This way, OpenAI will be able to see how the update has affected the software and its functions, and how users will adapt to the new changes, so that it can affect the wider audience at the right time.
Why This Matters for ChatGPT’s Free Tier
Usage limits appear to be among the most common ways for distinguishing between the free and paid subscriptions to ChatGPT. The ability to ask more questions or utilize the advanced functions without additional restrictions has prompted users to subscribe as continuing working with a lower-quality tool is frustrating.
The elimination of this limitation within the text-based conversation makes this option more attractive, removing this particular constraint for the majority of users, while other restrictions, such as limitations within image generation and voice input, still appear to be a decisive factor for some.
It can be argued that the users who asked questions or requested writing assistance comprise the vast majority within the target audience, which makes this update useful within the context of growth.
A Broader Pattern in AI Chatbot Competition
This update also comes at a moment when battle over attracting and retaining everyday users has become a key battleground for companies behind different AI chatbots. With products like ChatGPT increasingly being used by ordinary people to research, write and complete everyday tasks, firms have looked to make using their platforms by millions of free users as seamless as possible in order to ultimately drive more paying customers and ad revenue, depending on the business model.
OpenAI reaching 1 billion weekly users is a notable milestone in its growth. It would only be natural for them to want to reduce any impediments for their largest cohort of users, which will inevitably create more paying customers for the firm.
What This Means Going Forward
For everyday ChatGPT users, the most important changes to talk about this week are the transition to the new default model and the addition of the Think button this week and the introduction of unlimited text chats next week. The free and Go plans will see a reduction in disruptions during regular text-based conversations after the update has been fully implemented, as well as an increased ability to ask for more thorough reasoning on more difficult questions
Meanwhile, the move to the updated GPT-5.6 Sol model and the new sliding scale for plus and pro subscribers should give these users more control over how the assistant responds to different types of queries, which could be especially valuable for those who switch frequently between simple and multi-step requests.
Whether or not the reported reduction in hallucinations will have a noticeable effect on day-to-day use remains to be seen, as the improvements in the model’s factual accuracy have yet to be independently verified by external parties. However, the combination of unlimited text access for free users and improved reasoning controls across every tier represents one of the more substantial updates to ChatGPT’s free tier to date.
FAQs
Q1: What is the main change OpenAI announced for ChatGPT free users?
OpenAI is getting ready to unlock text chat limitations for all ChatGPT users, including those on the free plan.
Q2: What model will power the new unlimited free tier experience?
The new model of chatgpt which will serve as a default for free and go users is known as gpt-5.6 luna and it will replace gpt-5.5.
Q3: What is the new “Think” button?
A new “Think” feature is introduced for the free and go users which allow them to choose a higher level of reasoning when grappling with more complicated queries.
Q4: Are all features unlimited for free users now?
No, it does not mean everything is free because although the text chat will be free, there are still limits on files, images, voice, and image generation.
Q5: What changes are Plus and Pro users getting?
For the plus and pro accounts, they get access to an upgraded version of gpt-5.6 called gpt-5.6 ‘sol’ which is excellent for rapid response tasks like research, writing, and planning and a new thinking slider.
Q6: What does the thinking slider do?
The thinking slider allows the plus and pro users to control the amount of thought put into a query by the model based on how complex or involved the prompt is.
Q7: Is the updated GPT-5.6 Sol the same version used in Codex and Work?
It is not the same model as the one used in codex and work. The gpt-5.6 ‘sol’ that has been upgraded is different from the one used in OpenAI Codex and Work, which has remained the same.
Q8: Did OpenAI share any data on model accuracy improvements?
There were reports that internal evaluations at OpenAI reveal that GPT-5.6 Luna had 62% fewer factual errors than GPT-5.5-Instant, while GTP-5.6 Sol had 68% fewer errors.
