OpenAI is removing the daily wall that used to cut off free ChatGPT users mid conversation. Starting next week, anyone on the free tier or the low cost Go plan will be able to send as many text messages as they want, with no rate limit. Limits stay in place for photos, file uploads, voice mode, and image generation, so the change only covers plain typed conversation.
Free and Go users are also getting a new default engine called GPT-5.6 Luna, replacing the older GPT-5.5 model that had been standard since May. A new "Think" button lets these users ask the AI to spend more time reasoning through harder questions. Paying Plus and Pro subscribers get their own upgrade: a refreshed version of the flagship GPT-5.6 Sol model, plus a slider that lets them dial up or down how much thought goes into each answer.
OpenAI also claims a real accuracy jump behind these changes. The company said an internal evaluation found that, compared to GPT-5.5-Instant, factual errors were 62% less common for GPT-5.6 Luna and 68% less common for GPT-5.6 Sol. That matters for anyone using ChatGPT for real work, since wrong dates, numbers, or citations are the biggest reason business users hesitate to trust these tools. Still, this is OpenAI grading its own homework, not an independent test.
The timing here is not a coincidence. The Information has reported that OpenAI engineers found a way to cut the cost of running its models by more than half, at one point serving all logged out ChatGPT traffic with only a couple hundred graphics chips instead of tens of thousands. Free chats that used to be expensive to serve now cost OpenAI far less, which is very likely what made unlimited free texting affordable in the first place.
There is also a growth problem behind this generosity. OpenAI recently said its tools have crossed 1 billion weekly active users, a milestone the company had originally hoped to hit seven months earlier. Google's Gemini already has close to 950 million monthly users tied directly into its search engine, and Anthropic has pulled ahead of OpenAI in enterprise revenue. Giving free users a better product with fewer limits is a straightforward way to keep them from drifting to a rival chatbot.
There is a second motive worth watching closely. OpenAI began testing advertising inside ChatGPT for free and Go tier users earlier this year, placing sponsored content below chatbot answers. An assistant that people can chat with endlessly, without hitting a paywall, is also an assistant people spend more time inside of, which means more opportunities to serve ads and collect the data that makes those ads valuable.
For businesses, the practical takeaway is simple. Free AI tools are getting genuinely more capable and less limited, which is good news if your team relies on the free tier for quick research or drafting. But the underlying business model is shifting toward advertising, the same model that shaped search engines and social media. Anyone building workflows around a free AI assistant should expect that assistant to eventually come with sponsored answers attached, and should treat the free tier as a media product, not a neutral tool.