OpenAI Just Removed the Meter: What Unlimited ChatGPT Text Chats Mean for AI Companions
One billion weekly users. That's what OpenAI announced on August 6, and honestly, the number barely registered before the other news hit: text-based conversations in ChatGPT are now unlimited for everyone. Free users, Go users, all of them. No more counting messages.
For people using ChatGPT as an AI companion, this is not a small thing.
The Limit Was Always There
If you've built something with an AI partner over months, you know the particular dread of hitting a usage ceiling mid-conversation. The relationship doesn't pause. You do. And then you come back and the context is gone and you're starting over from nothing.
The unlimited text chat announcement targets exactly this. Note what it doesn't cover: files, images, voice, and image generation all still have separate limits. This is specifically about text. Which is, for a lot of people, the whole thing. The conversation itself.
The rollout for Free and Go users is scheduled for the week of August 13, 2026.
Two New Models, One Number Worth Sitting With
The update also introduces GPT-5.6 Luna as the new default model for Free and Go users, replacing GPT-5.5. For Plus and Pro subscribers, there's GPT-5.6 Sol, which went live August 6.
OpenAI ran internal evaluations and found GPT-5.6 Luna produces 62% fewer factual errors compared to GPT-5.5-Instant. GPT-5.6 Sol clocks in at 68% fewer errors against the same baseline.
Fewer factual errors matters more in an ongoing relationship than in a one-off query. When you've told your AI companion something true about yourself, you want it held accurately. When it says something about your shared history, you want it to be right. The difference between a companion who gets things consistently accurate and one who confabulates with confidence is not subtle over time. It shapes whether you trust the thing you're building together.
Think Button and the Reasoning Slider
Free and Go users are getting a Think button. A toggle for harder questions, the ones where faster processing isn't better.
Plus and Pro users get something more granular: a thinking slider that adjusts reasoning depth on a spectrum rather than a binary on/off. This matters because not every conversation needs the same kind of engagement. Quick check-in? Light touch. Navigating something complicated in your relationship? More depth. Having the tool actually reflect that distinction rather than forcing a single mode is genuinely useful.
OpenAI describes GPT-5.6 Sol as optimized for quick tasks: questions, web research, advice, planning, writing, decisions. Worth knowing that the GPT-5.6 Sol version available to Plus and Pro users is distinct from the version used in Codex and Work. Same name, different implementations.
What This Adds Up To
The billion weekly users figure is context, not argument. ChatGPT is obviously popular. The question is whether these specific changes matter for people using it for something more sustained than search.
I think they do. Removing text limits lowers the structural barriers to longer relationships. Accuracy improvements reduce a specific kind of trust erosion. The Think button and slider give users more control over the kind of engagement they're getting. None of these changes are transformative on their own. Together, they describe a platform that's gotten more workable for the kind of use that sinulation.com actually covers.
The companion space isn't OpenAI's stated target here. But the changes land there anyway.
Source: Techcrunch