OpenAI has removed limits on text-based chats for every ChatGPT user, a move that arrives just after the chatbot crossed 1 billion weekly users. The company also introduced two new models, GPT-5.6 Luna and GPT-5.6 Sol, along with fresh controls for how much reasoning the AI applies to a question.
New default model for Free and Go users
GPT-5.6 Luna becomes the default model for Free and Go users, replacing GPT-5.5. These users will also get a new "Think" button that unlocks higher reasoning power for tougher queries. OpenAI said separate limits will remain for files, images, voice, and image generation, so the removal of caps applies only to text conversations.
Other changes for Free and Go users are arriving this week. Next week, the company will deliver unlimited text chats and the Think button for harder questions, completing the rollout for the lower tiers.
Upgraded model for Plus and Pro subscribers
Plus and Pro users gain access to the upgraded GPT-5.6 Sol model, which OpenAI says is better for quick tasks such as questions, web research, advice, planning, writing, and decisions. The new Sol gives more compact and robust answers, according to the company. A separate version of Sol for Codex and Work remains unchanged.
Plus and Pro users also get a thinking slider to adjust how much thought the model puts into an answer. The slider can be tuned based on the complexity and steps involved in a given request. The updated GPT-5.6 Sol is available to Plus and Pro users today.
Fewer factual errors in internal tests
OpenAI said internal evaluation found factual errors 62% less common for GPT-5.6 Luna compared with GPT-5.5-Instant. For GPT-5.6 Sol, factual errors were 68% less common against the same baseline. The company presented these figures as evidence of improved reliability across both tiers.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
The reductions mark a notable step forward in accuracy, though OpenAI framed them as internal findings rather than independent results. The previous model, GPT-5.5, served as the reference point for the free tier comparison.
Broader context and editorial notes
The announcement comes from TechCrunch reporter Ivan Mehta, who is based in India and previously worked at Huffington Post and The Next Web. The article includes image credits to OpenAI and links to purchase options that may earn commission. Mehta can be reached at im@ivanmehta.com or on Signal at ivan.42.
TechCrunch Disrupt takes place October 13-15 in San Francisco, with early registration saving up to $330. The event will feature speakers and panels across the tech industry.
Other stories in the same editorial roundup include Bending Spoons' plan to buy Airtable for $1.28 billion, Sequoia leading a $1 billion funding round for Valar Atomics, backlash from influencers over an OpenAI luxury trip, Malaysia shutting down Network School, Hank Green's use of AI, fraud among VC-backed startups, and WhatsApp testing a folder for business messages.
What this means for users
The removal of text chat limits signals a shift in how OpenAI treats the free tier, which previously restricted how many messages users could send. With 1 billion weekly users, the change could affect a massive audience. The new Think button and thinking slider give users more control over the trade-off between speed and depth in responses.
OpenAI has not said whether the unlimited text chats will apply to all future models or only the current generation. The company also did not specify any changes to pricing for Plus or Pro tiers. For now, the updates roll out in stages, with the Sol model live today and the remaining Free and Go features arriving over the next week.