OpenAI is quietly erasing one of ChatGPT’s most persistent annoyances. Starting next week, free and Go tier users will get unlimited text chats, ditching the rate limits that currently cut conversations short. If you’ve ever been mid-brainstorm only to hit a “you’ve reached your limit” wall, that friction disappears. Uploads and image-based messages will still face caps, but pure text is now a free-for-all.
Alongside the unlimited access, free users get a new “Think” button for tougher questions and a default model upgrade to GPT-5.6 Luna this week. It’s a clear signal: OpenAI wants the free experience to feel less like a trial and more like a viable product in its own right. The gap between free and paid is narrowing, but OpenAI is also giving subscribers new reasons to pay up.
For Plus and Pro users, GPT-5.6 Sol is getting a fact-reliability overhaul starting Thursday. OpenAI says the model is “designed to make fewer mistakes — especially when answers depend on dates, numbers, sources, rules, or assumptions.” It will also deliver tighter, more direct responses and skip unhelpful detail. That’s a direct answer to the persistent hallucination problem that plagues even the best models. Plus and Pro subscribers also get a new slider to control how much compute ChatGPT dedicates to each answer — think of it as a dial between speed and depth.
The dual-track approach is fascinating. Free users get breadth — unlimited chats, a better default model, and reasoning access. Paying users get depth — more reliable facts, tighter answers, and compute control. It’s the freemium playbook executed with unusual clarity. The question is whether “fewer mistakes” translates to meaningfully fewer hallucinations in practice, or if it’s just another reliability promise in an industry full of them.
