The users of ChatGPT who use a free account will be able to maintain text conversations without worrying about reaching the message limit coming soon. OpenAI has announced a significant expansion of its service that also affects the plan Go and that will arrive accompanied by a new default model and an option to resort to more advanced reasoning when necessary.
The most important change to the plans Free and Go will be the disappearance of limits on text chats starting the week of August 10. Until now, ChatGPT restricted the number of messages that could be sent during a certain period before temporarily limiting access. Removing this barrier It will only affect text, so there will still be limits on uploading files, generating images and using other tools.
OpenAI will also convert this week GPT-5.6 Moon in the default model for Free and Go users to replace GPT-5.5 Instant. Luna is part of the new family GPT-5.6 and it is the fastest and most economical model of the three that comprise it. Above are Terraconceived as an intermediate option, and Sunthe flagship and most powerful model of the family.
Free and Go users will also receive a new button during the week of August 10 ‘Think’. When activated, GPT-5.6 Luna will have more reasoning time to address complex questions. OpenAI presents it as a way to ‘access a higher level of reasoning for more difficult questions.’
The accounts Plus and Pro They receive other improvements that already began to be deployed this Thursday. OpenAI has updated GPT-5.6 Sol to make it the model used both in instant responses and when ChatGPT puts more thought into a query. The company assures that it now offers ‘more direct’ answers, uses a more concise format and avoid adding information that is not useful.
The accuracy of GPT-5.6 Sol has also been one of the main objectives of this update. OpenAI ensures that the model is now ‘more reliable with facts’ and is designed to commit fewer errors when an answer depends on dates, figures, sources, standards or assumptions. In its internal tests with financial, medical and legal consultations, responses containing at least one factual error were 68% less frequent with the new Sol than with GPT-5.5 Instant. On GPT-5.6 Luna, the reduction reached 62%.
Since yesterday, Plus and Pro subscribers have a new slider to decide how much reasoning ChatGPT should put into each response. The feature is available on the web and in mobile and desktop applications, so the user can prioritize speed in simple questions or increase reasoning for tasks such as researching, writing, programming or planning.
The GPT-5.6 Sol update is limited to conventional ChatGPT chats for now. OpenAI has clarified that the model versions used in ChatGPT Work and Codex They do not change with this update.