OpenAI removes ChatGPT text limits for free users and adds new reasoning controls

OpenAI is expanding ChatGPT access by removing limits on text-based chats for Free and Go users, a notable change for a product the company says recently passed 1 billion weekly users. The update is paired with new model options and controls designed to give users more flexibility over how much reasoning power ChatGPT applies to a response.

Unlimited text chat for Free and Go

According to OpenAI, Free and Go users will get unlimited text chats powered by a new default model, GPT-5.6 Luna, which replaces GPT-5.5 for those tiers. The company is also introducing a new Think button for complex questions, allowing users to invoke higher reasoning effort when needed.

OpenAI said the unlimited access applies to text conversations only. Separate usage limits will still remain in place for files, images, voice, and image generation.

New options for Plus and Pro subscribers

Paid ChatGPT tiers are also getting changes. Plus and Pro users will receive an upgraded GPT-5.6 Sol model aimed at fast everyday tasks such as web research, planning, writing, advice, and decision support. OpenAI says the updated Sol model is designed to produce more compact and robust answers.

In addition, Plus and Pro users are getting a thinking slider, which lets them adjust how much effort the model spends on a response. That gives subscribers a more granular alternative to a single high-reasoning mode, and reflects a broader trend in AI products toward user-controlled tradeoffs between speed and depth.

Why this matters

The biggest headline is the removal of text-chat limits for free users. Consumer AI services have typically used message caps as a core way to manage computing costs and differentiate paid plans. By lifting those limits for basic text usage while keeping caps on more expensive features like voice and image generation, OpenAI appears to be narrowing the gap between free and paid access without fully giving up tiered monetization.

The new reasoning controls are also significant. A simple Think button for lower tiers and a more configurable thinking slider for paid users suggest OpenAI is making model effort a visible product feature, rather than an invisible backend choice. That could help users better match ChatGPT's behavior to different tasks, from quick lookups to multi-step problem solving.

Performance claims and rollout

OpenAI said internal testing showed factual errors were 62% less common with GPT-5.6 Luna and 68% less common with GPT-5.6 Sol, compared with GPT-5.5-Instant. As with all vendor-provided benchmarks, those figures should be viewed as directional until independently validated.

The upgraded GPT-5.6 Sol model for Plus and Pro users is available immediately, while the Free and Go changes are rolling out this week. OpenAI said unlimited text chats and the new Think button for those users will become available next week.

The broader picture

The update underscores how competitive the mainstream AI assistant market has become. Product differentiation is increasingly shifting from raw model access toward usability features, reliability improvements, and controls that let users decide when they want fast answers versus deeper reasoning. For OpenAI, making unlimited text chat available on free tiers could help lock in habit and scale, while reserving premium features for more compute-intensive workflows.