OpenAI will remove text message limits for free ChatGPT accounts starting the week of August 10, allowing users to hold text conversations without message caps. The expansion applies to both Free and Go plans as part of a series of updates to the service.
Until now, ChatGPT restricted the number of text messages a user could send over a set timeframe before temporarily limiting access. Removing this barrier applies only to text chats, meaning limits remain for uploading files, generating images, and using other platform tools.
OpenAI is also making GPT-5.6 Luna the default model for Free and Go accounts this week, replacing GPT-5.5 Instant. Luna is part of the new GPT-5.6 family and is the fastest and most economical model among the three that make up the line.
The GPT-5.6 model family features three distinct tiers. Terra is designed as an intermediate option, while Sol stands as the flagship and most powerful model in the lineup.
New reasoning controls for free accounts
During the week of August 10, Free and Go users will also receive a new button labeled Think. When activated, the button gives GPT-5.6 Luna additional reasoning time to handle complex questions.
OpenAI presented the feature as a way for free users to access a higher level of reasoning when tackling more difficult questions.
Upgrades for paid Plus and Pro subscribers
Subscribers on paid Plus and Pro plans began receiving separate improvements on Thursday. OpenAI updated GPT-5.6 Sol so that it serves as the active model for both instant responses and queries where ChatGPT applies extra reasoning.
The company stated that the updated Sol model now delivers more direct answers in a more concise format while avoiding unhelpful information. OpenAI made factual accuracy a central focus of the update, designing Sol to make fewer mistakes when answers depend on dates, numbers, sources, rules, or assumptions.
In internal tests covering financial, medical, and legal queries, responses with at least one factual error occurred 68 percent less often on the new Sol model than on GPT-5.5 Instant. On the GPT-5.6 Luna model, factual errors dropped by 62 percent.
Paid Plus and Pro users also gained access on Thursday to a new reasoning slider control across web, mobile, and desktop applications. The slider allows users to set how much reasoning time ChatGPT devotes to a query, letting them prioritize speed for simple questions or increase reasoning for research, writing, programming, or planning.
Scope of the model updates
The rollout of the updated GPT-5.6 Sol model is currently limited to conventional ChatGPT chats. OpenAI clarified that versions of the model used in ChatGPT Work and Codex do not change with this update.
