OpenAI is lifting the message cap on ChatGPT’s free tier, the company said August 6, giving unlimited text chats to users who previously had to ration prompts on its flagship models.
Free and Go-tier users had been capped at roughly 10 to 40 messages every three to five hours before being switched to a lighter model for the rest of the window. That limit disappears as GPT-5.6 Luna becomes the default model for both tiers; unlimited text chats and a new “Think” button, which gives Luna extra time to work through harder questions, are rolling out during the week of August 10, according to OpenAI.
Fewer errors, but still limited elsewhere
OpenAI said internal evaluations found GPT-5.6 Luna produced a factual error 62% less often than its predecessor, GPT-5.5 Instant. The unlimited offer covers text only — free users still face limits on file uploads, image generation and other tools, and OpenAI said access remains subject to abuse guardrails.
Paying subscribers get a parallel upgrade. GPT-5.6 Sol, the flagship model for Plus and Pro users, has been retuned for faster, more focused answers on everyday tasks like planning, research and writing; OpenAI cited a 68% drop in factual errors compared with GPT-5.5 Instant. Subscribers can also adjust a new reasoning-depth slider to trade speed for deeper analysis.
The change follows OpenAI’s disclosure that ChatGPT has reached roughly 1 billion weekly active users, and lands weeks after the company cut GPT-5.6 Luna’s price 80% and had Sol rewrite its own GPU serving code to cut inference costs. Together, the moves point to the same approach: route free-tier demand to a cheap, fast model and save heavier compute for paying customers.
Dropping rate limits also raises the stakes in the consumer chatbot market, where OpenAI competes with Google’s Gemini and Anthropic’s Claude for casual users who might otherwise abandon a conversation mid-way. OpenAI has not said whether the change will extend to ChatGPT’s API or enterprise tiers.