Skip to main content

ChatGPT Free Users Get Unlimited Chat with GPT-5.6 Luna

OpenAI just dropped a big update for ChatGPT, and it's good news for everyone—especially those on the free tier. Starting this week, free users get unlimited text conversations, powered by the new default model, GPT-5.6 Luna. No more hitting a wall after a few exchanges; you can chat as much as you want, whether you're brainstorming ideas, planning a trip, or just asking random questions.

But that's not all. Plus and Pro subscribers are also getting some love. Their ChatGPT chat mode now runs on GPT-5.6 Sol, which has been specially optimized for conversational quality. According to OpenAI, this version shows marked improvements in factual accuracy and answer quality. Sam Altman himself took to X (formerly Twitter) to hype it up, saying, "5.6 Sol performs much better in chats, and free users can also have unlimited text chats!"

So, what's the big deal with these new models? OpenAI says a billion people use ChatGPT every week for quick Q&A, web searches, planning, research, consulting, and even complex decision-making. The goal with GPT-5.6 Sol is to handle all of that smoothly. The model now gives more focused responses, adjusting the level of detail based on what you ask. It avoids unnecessary formatting and even offers helpful corrections when you're just agreeing without adding value. For Plus and Pro users, the same model powers both instant answers and deep reasoning, so the experience feels more consistent.

Take a simple example: if you ask about cycling conditions, the older GPT-5.5 Instant might list every detail—rain, wind, temperature—all at once. But GPT-5.6 Sol zeroes in on what actually matters to you. It identifies wind as the key factor, skips the irrelevant stuff, and even generates a handy weather forecast sheet. That's the kind of smart, context-aware response that makes you go, "Ah, that's exactly what I needed."

One of the most impressive improvements is in reducing hallucinations. OpenAI's internal tests show that for financial, medical, and legal prompts, the rate of at least one factual error dropped by about 62% with GPT-5.6 Luna compared to GPT-5.5 Instant, and by about 68% with GPT-5.6 Sol. That's a huge leap in reliability, especially for users who depend on ChatGPT for accurate information.

Subscribers also get a slick new UI tweak. Remember the hassle of selecting both the model and inference strength separately? That's gone. The chat interface now includes a slider, similar to what you see in Codex. Plus and Pro users can adjust how much the model "thinks" before answering—keep it in fast mode for everyday questions, or crank it up when you're planning, researching, writing, coding, or making big decisions. It's available on web, mobile, and desktop.

For free users, the unlimited chat is a game-changer, but there are still some limits. Image generation and file uploads remain capped. Also, since Luna isn't a universal model, OpenAI added a new "Think" button to handle complex issues that need deeper reasoning.

One thing to note: this update is specifically for the ChatGPT chat experience. The GPT-5.6 Sol version used in Work and Codex isn't part of this rollout. So, if you're using those tools, you'll have to wait for their own updates.

All in all, this update makes ChatGPT more accessible and smarter for everyone. Whether you're a free user enjoying unlimited chats or a subscriber tweaking the thinking slider, there's something to be excited about. It feels like OpenAI is really listening to what users want—more freedom, better answers, and less fluff.

Key Points

  • Free ChatGPT users now get unlimited text chats with GPT-5.6 Luna, rolling out this week.
  • Plus and Pro users get GPT-5.6 Sol, optimized for better factual accuracy and answer quality.
  • Hallucination rates dropped by ~62% (Luna) and ~68% (Sol) in financial, medical, and legal prompts.
  • New slider in Chat lets Plus/Pro users control thinking depth for different tasks.
  • Free users still have limits on image generation and file uploads; a "Think" button handles complex queries.
  • The update applies only to ChatGPT chat, not Work or Codex.