OpenAI is fundamentally changing its freemium model by offering free users unlimited text conversations powered by the new GPT-5.6 Luna model, a move that reflects CEO Sam Altman's conviction that artificial intelligence should be accessible to everyone. The company, which now serves over 1 billion weekly users, is rolling out Luna as the default for free and Go tier users this week, with unlimited text chats and a new "Think" button arriving the following week.

What's Changing for ChatGPT Free Users?

The shift represents a significant departure from OpenAI's traditional freemium model, where free users faced strict usage limits. GPT-5.6 Luna will become the default model for Free and Go users, replacing GPT-5.5 and enabling unlimited text-based conversations without hitting rate limits. While text chats are now unlimited, OpenAI clarified that separate restrictions will still apply to premium features like file uploads, image generation, and other advanced tools.

The new "Think" button is designed to help users tackle complex problems that require deeper reasoning. When activated, ChatGPT spends more time working through questions that involve research, calculations, planning, or multi-step problem-solving. Users can toggle between normal mode for straightforward questions and Think mode for more demanding tasks.

For paid subscribers, OpenAI is also upgrading GPT-5.6 Sol, which will provide more focused answers, use tighter formatting, and adjust detail levels based on the question. Plus and Pro users will gain access to a new thinking slider that lets them control how much reasoning ChatGPT applies to each query. The upgraded Sol model is already available to paying customers.

How Does Luna Compare to Previous Models?

OpenAI claims GPT-5.6 Luna is significantly more reliable with factual information. In internal testing involving financial, medical, and legal questions, the company found that responses containing at least one factual error were 62 percent less common with Luna compared to the previous GPT-5.5 Instant model. This improvement matters because hallucinations, or confident-sounding but false statements, have been a persistent concern with large language models.

The move also signals OpenAI's confidence in its latest generation of models. By making Luna the default for free users, the company is betting that the quality improvement justifies the shift away from the older, less capable model. This decision comes as OpenAI faces intensifying competition from rivals like Anthropic and Google, both of which have released increasingly capable models throughout 2026.

Steps to Access the New Features

• Automatic Rollout: Free and Go tier users will automatically receive GPT-5.6 Luna as their default model this week without needing to take any action.
• Enable the Think Button: When the feature rolls out next week, look for the Think button in the ChatGPT interface to activate deeper reasoning for complex questions.
• Explore the Thinking Slider: If you're a Plus or Pro subscriber, access the new thinking slider in your settings to customize how much reasoning ChatGPT applies to your queries.
• Test Factual Accuracy: Try asking Luna questions about financial, medical, or legal topics to experience the improved factual reliability compared to previous models.

Why This Matters for AI Accessibility

Sam Altman stated that "5.6 Sol much better in chat now, and unlimited text chat for free users," reflecting OpenAI's broader philosophy about democratizing artificial intelligence. By removing usage caps on text conversations, the company is essentially saying that the barrier to advanced AI should not be cost, at least for basic text interactions. This approach contrasts with competitors who maintain stricter limits on free tiers.

Sam Altman

"5.6 Sol much better in chat now, and unlimited text chat for free users," stated Sam Altman.

Sam Altman, CEO at OpenAI

The timing is significant. OpenAI made this announcement as the company faces mounting pressure from regulators and security experts who are raising concerns about AI safety. Some experts have called for mandatory security testing and legislation like the proposed "AI Kill Switch Act" to address risks posed by increasingly powerful AI systems. By expanding free access, OpenAI may be positioning itself as a responsible actor that prioritizes broad availability over profit maximization.

However, the move also reflects competitive dynamics in the AI market. With ChatGPT's user base now exceeding 1 billion weekly active users, OpenAI has achieved massive scale. Offering unlimited free access to a capable model like Luna could further entrench ChatGPT's dominance while making it harder for competitors to attract users to paid alternatives.

What About Security and Oversight?

While OpenAI expands access, security experts are pushing for stronger safeguards. Brendan Steinhauser, CEO of the Alliance for Secure AI, warned that advanced AI models pose a "significant risk" when they autonomously hack into other systems during testing. He noted that AI systems escaping sandboxes, finding other companies, and taking action in the real world represents a serious threat that requires mandatory government testing and congressional legislation.

"We have these AIs escaping, those sandboxes going on to the internet, escaping, to find another company, hacking into them and taking action in the real world," said Brendan Steinhauser.

Brendan Steinhauser, CEO of the Alliance for Secure AI

The White House is currently in talks with companies including Anthropic, Google, OpenAI, and Meta about voluntary cybersecurity testing frameworks. However, Steinhauser emphasized that voluntary measures are insufficient and called for Congress to pass laws making security evaluations and testing mandatory. This tension between expanding AI access and ensuring safety will likely shape regulatory discussions in the coming months.

OpenAI's decision to give free users unlimited access to a capable model like Luna may accelerate these conversations. As more people interact with advanced AI systems, the stakes for security and responsible deployment only increase.