OpenAI updates ChatGPT with unlimited text for free users and smarter models

OpenAI has announced a significant update to its ChatGPT platform, introducing unlimited text-based conversations for users on its Free and Go tiers. Starting next week, these users will no longer face text-based rate limits, though restrictions on file uploads, image generation, and other specialized tools remain in place. As part of this transition, the GPT-5.6 Luna model is becoming the default for these users, replacing the previous GPT-5.5 Instant model.

For subscribers on the Plus and Pro tiers, OpenAI is rolling out an updated version of the GPT-5.6 Sol model. This iteration is specifically tuned for everyday conversation, aiming to provide more focused responses by dynamically adjusting the level of detail based on the user's query. The company stated that the model is designed to avoid unnecessary formatting and provide direct answers, which it claims will improve clarity for both quick questions and complex tasks like multi-step planning or research.

The update also introduces a new reasoning slider for Plus and Pro users, available across web, mobile, and desktop applications. This control allows users to manually adjust the amount of thought the model applies to a response, ranging from quick, concise answers to more intensive processing for coding or analytical tasks. OpenAI reports that this version of GPT-5.6 Sol is more reliable when handling prompts involving dates, numbers, rules, and assumptions.

Internal evaluations conducted by OpenAI suggest that the updated GPT-5.6 Sol model demonstrates improved factual accuracy. According to the company, responses containing at least one factual error were 68% less common when using the new model compared to the previous GPT-5.5 Instant default, based on testing across financial, medical, and legal prompts. These improvements are currently limited to the standard chat experience, as the versions of GPT-5.6 Sol used in ChatGPT Work and Codex remain unchanged.

Free and Go users will also gain access to a new Think button starting next week, which is designed to assist with more complex queries by allowing the model to dedicate additional reasoning time to the response. While the GPT-5.6 Luna model is being deployed as the new default for these users this week, the full suite of new features, including the unlimited text capability and the Think button, will be available by the following week.