OpenAI is removing limits on text-based chats for every ChatGPT user, the company said this week, days after ChatGPT crossed 1 billion weekly users.

The change centers on a new model called GPT-5.6 Luna, which becomes the default for Free and Go tier users, replacing GPT-5.5. Free and Go users are also getting a new “Think” button that lets them request higher reasoning effort on complex questions, though OpenAI said separate limits will remain in place for files, images, voice and image generation.
Free ChatGPT users have long faced a daily cap on messages before being routed to a lighter fallback model, a restriction OpenAI introduced as compute costs climbed alongside its user base. Removing that cap for text chats marks one of the more generous free-tier changes the company has made since ChatGPT launched.
Paying Plus and Pro subscribers are getting a different upgrade: access to an improved GPT-5.6 Sol model built for faster tasks like research, writing, planning and everyday decisions. OpenAI said this version of Sol is separate from the Sol variant used inside Codex and Work products, which is unchanged.
Plus and Pro users also gain a thinking slider, letting them dial reasoning effort up or down depending on how complex a question is, rather than relying on a fixed setting.
OpenAI said internal testing found factual errors dropped 62% for GPT-5.6 Luna and 68% for GPT-5.6 Sol, compared with the outgoing GPT-5.5-Instant model.
The rollout is staggered. The upgraded Sol model is already live for Plus and Pro users. The Luna model and Think button are arriving for Free and Go users this week, with unlimited text chats following next week.
OpenAI has not said whether the unlimited-chat policy extends to voice conversations or the ChatGPT mobile app, where usage patterns and server costs differ from the web version.
The update lands as OpenAI faces growing competition from Google’s Gemini and Anthropic’s Claude, both of which have pushed their own free-tier improvements in recent months to hold on to casual users who might otherwise drift to a rival chatbot.



