OpenAI says ChatGPT Free and Go users will receive unlimited text chats next week, using its GPT-5.6 Luna model. The change removes the stated cap for plain conversations, though it comes with abuse guardrails and does not make every part of ChatGPT unlimited.
File uploads, images and other tools will still be subject to limits. That distinction is the fine print worth reading: a user can keep typing to ChatGPT, but attaching documents or using non-text features remains governed by separate caps.
OpenAI announced the changes on August 6. It said Luna will become the default model for Free and Go users during the announcement week; unlimited text chats and the new Think button are scheduled for the following week.
What will remain limited for ChatGPT Free users?
OpenAI says the unlimited allowance applies only to text chats. Limits remain on file uploads, images and other tools, and unlimited text access is subject to the company’s abuse guardrails.
ChatGPT Go is OpenAI’s lower-cost paid subscription. It includes the Free plan along with expanded access to core models and tools including image generation, file uploads and advanced data analysis, according to OpenAI’s help documentation. Go usage also remains subject to plan limits, guardrails and possible restrictions during high demand.
A Think button for harder prompts
Free and Go users will also get a Think button, a control for harder questions. OpenAI says tapping it gives GPT-5.6 Luna more time to work through an answer. In practical terms, the company is offering a way to request more reasoning effort without changing the scope of the text-chat promise.
Luna is the model becoming the default on the two tiers. OpenAI said its internal evaluation of financial, medical and legal prompts found responses containing at least one factual error were about 62% less common with Luna than with GPT-5.5 Instant. That is a company-run test, not an independent verification of the result, and users should still check consequential claims. Large language models generate text from patterns in context; an answer that sounds settled is not automatically a reliable source.
Paid subscribers get a separate update
The Free and Go rollout is separate from changes for Plus and Pro subscribers. Those users can access an updated GPT-5.6 Sol in ChatGPT starting August 6, plus a slider on web, mobile and desktop that lets them choose how much thought the system puts into an answer.
OpenAI says the revised Sol is intended to provide more focused responses and make fewer factual mistakes. In the same internal evaluation, the company said responses with at least one factual error were about 68% less common for Sol than for GPT-5.5 Instant. The Sol update applies to ChatGPT’s Chat experience; OpenAI says the Sol versions used by Work and Codex are unchanged.
This story draws on original reporting from The Verge.
