SAN FRANCISCO, CALIFORNIA / RankWire.AI / – OpenAI has announced the removal of text chat restrictions for its Free and Go plans by adopting GPT-5.6 Luna as the default model, with the company indicating that unlimited text conversations will commence next week, while safeguards against misuse remain in place. During this rollout, Luna also becomes the default model for these plans, and a new Think button will be available to Free users at the same time, enabling the model to dedicate additional time for reasoning through more challenging questions before providing an answer.
This expanded allowance pertains exclusively to standard text interactions and does not encompass other features within ChatGPT, such as image generation, file uploads, or tools that require more extensive computational resources. OpenAI stated that the Think button will be introduced alongside the unlimited text access next week, allowing users to activate it when a question benefits from extra reasoning time. This update broadens the core conversational capabilities for Free and Go accounts, while maintaining separate limits for other functionalities and services.
GPT-5.6 comprises multiple models, including Luna, Sol, and Terra, each tailored to different performance needs. Luna is designed to be the faster and more cost-effective option within this family. According to OpenAI, approximately one billion people now use ChatGPT weekly for activities such as research, planning, writing, and searching. With this latest update, Luna becomes the primary ChatGPT model for Free and Go users, while other GPT-5.6 variants remain accessible within different products and plans suited for various workloads and access levels.
Expanded free access while tool usage limits stay in effect
Subscribers to ChatGPT Plus and Pro are now receiving an upgraded GPT-5.6 Sol model for both routine tasks and more complex reasoning, with OpenAI noting that Sol now provides more targeted answers and minimizes unnecessary detail when shorter responses suffice. Paid members also gain a reasoning slider compatible with web, mobile, and desktop platforms, which allows them to control the level of processing the model applies to each request, useful for tasks such as research, coding, writing, and planning that require varying degrees of reasoning.
OpenAI additionally shared results from internal tests assessing the factual accuracy of its newest models, focusing on responses to financial, medical, and legal questions that demand detailed factual correctness. The company reported that GPT-5.6 Luna produced approximately 62% fewer responses with at least one factual error compared to GPT-5.5 Instant. Similarly, GPT-5.6 Sol demonstrated a reduction of about 68% under the same evaluation criteria. The company also highlighted improvements in handling dates, figures, sources, rules, and assumptions when these details influenced the accuracy of the responses.
New models incorporate safeguards for younger users
The GPT-5.6 series introduces enhanced safeguards aimed at users under 18, including restrictions on romantic roleplay, age-restricted challenges, and claims that might promote replacing real-world relationships. Additional boundaries are set against sexual content, eating disorders, dangerous activities, restricted substances, and graphic violence. OpenAI stated that it has implemented further assessments for responses involving younger users, and these same abuse safeguards will remain in effect when unlimited text conversations are available to Free and Go accounts through the expanded ChatGPT access.
This rollout applies to the primary ChatGPT service and does not impact all products utilizing GPT-5.6. Versions such as GPT-5.6 Sol used within ChatGPT Work and Codex remain unaffected by this update. The updated Sol model and reasoning controls are now accessible to Plus and Pro subscribers, while Free and Go users will see Luna as their default model this week. The unlimited text conversations and Think button will be introduced next week, with separate limits still in place for images, uploads, and other tools requiring significant processing power.