OpenAI just made ChatGPT free for unlimited use — and most people have no idea what changed. As of August 7, 2026, free-tier users can have unlimited text conversations with GPT-5.6 Luna, no daily caps, no throttling, no asterisks. It’s the most significant change to ChatGPT’s free plan since the product launched, and it comes with a smarter model, a new “Think” button, and a trade-off that power users will want to know about.
The update landed on August 6 at roughly 5pm PT, tucked inside a broader announcement about GPT-5.6 Sol getting unified for paid users. But the free-tier expansion is the bigger story for most people. Here’s exactly what changed, what you get, and what to watch out for.
What Changed for Free Users
Previously, free-tier ChatGPT users faced daily message caps on GPT-4-era models and had to wait for GPT-5.6 access. Starting August 7, the default model for free users switched to GPT-5.6 Luna — OpenAI’s most capable free-tier model to date — with unlimited text chats. That means you can fire off as many text prompts as you want without hitting a wall.
There’s one important scope limit: the unlimited access applies to text chat only. Voice conversations, image generation, and other multimodal features remain separate. If you were expecting unlimited image generation for free, that’s not what shipped here. But for anyone who uses ChatGPT primarily for writing, research, brainstorming, coding help, or analysis — this is the upgrade that matters.
The “Think” Button
Free users now also get a “Think” button that triggers deeper reasoning on specific questions. It’s the free-tier version of the reasoning slider that Plus and Pro subscribers get. Hit “Think” before asking a complex math problem, a multi-step coding question, or anything that benefits from deliberate, chain-of-thought reasoning. It’s not always on — you choose when to activate it — which keeps response times snappy for simple queries while giving you on-demand depth.
What Changed for Paid Users (Plus and Pro)
For Plus ($20/month) and Pro ($200/month) subscribers, the update merged ChatGPT’s separate “Instant” and “reasoning” models into a single GPT-5.6 Sol experience. Previously, ChatGPT routed quick questions to a fast model and complex questions to a slower reasoning model — two separate systems. Now, one model handles both, with a reasoning-effort slider letting you dial how much thinking power to apply.
OpenAI claims the updated GPT-5.6 Sol produces 68% fewer factual errors than the previous GPT-5.5 Instant model on finance, medicine, and law benchmarks. Take that number with a grain of salt — it’s an OpenAI-internal benchmark, not an independent audit — but the direction is clear: the paid model got more reliable.
The Model Picker Is Gone (Sort Of)
The most controversial part of this update is what disappeared. The manual model picker — the dropdown that let you switch between GPT-4, GPT-4o, GPT-5.5, and other models mid-conversation — has been pushed into Settings for most users. OpenAI’s position is that most people shouldn’t need to choose a model manually; the system should route intelligently on its own.
Power users on X have been vocal about the friction this creates. If you’re someone who switches models mid-conversation for different task types — say, GPT-5.5 for creative writing and GPT-5.6 Sol for technical analysis — you’ll now need to dig into Settings each time. It’s a usability regression for a small but vocal minority, and OpenAI may course-correct if the pushback continues.
How to Get the Most Out of Free ChatGPT
Here’s how to make the most of the new free tier:
- Use the “Think” button strategically. Don’t activate it for simple questions like “what’s the weather” — it adds latency. Save it for coding problems, data analysis, multi-step research, and anything where accuracy matters more than speed.
- Leverage unlimited text for bulk tasks. Brainstorming sessions, iterative writing, code debugging — these all benefit from being able to ask 50 follow-up questions without worrying about hitting a cap.
- Treat GPT-5.6 Luna as a different tool than Sol. Luna is faster and less expensive to run, which is why OpenAI can offer it unlimited. It’s genuinely capable for most tasks, but for the absolute highest-quality outputs on complex reasoning, Sol (paid) still has an edge.
- Test the factuality yourself. OpenAI’s 68% improvement claim is based on their own benchmarks. If you use ChatGPT for medical, legal, or financial information, always verify critical claims independently. The model is better — not perfect.
Why This Matters
ChatGPT hitting roughly 1 billion weekly active users changes the calculus for OpenAI. The free tier is no longer a teaser — it’s the product. By making text chat unlimited and upgrading the underlying model, OpenAI is betting that free users will convert to paid plans for voice, image generation, and the more capable Sol model over time. It’s the same playbook Spotify and Dropbox used: make the free tier genuinely good, then let the premium tier sell itself.
For the average person who uses ChatGPT a few times a week to write emails, explain concepts, or brainstorm ideas, the free plan just became the only plan you need. And that’s a bigger deal than any model benchmark number.
The bottom line: If you’re on the free tier, you now have unlimited text conversations with a model that’s meaningfully better than what you had last month. Use it. If you’re on Plus or Pro, the unified Sol experience with the reasoning slider is a genuine upgrade. The model picker change is annoying but manageable. Either way, August 2026 is the month ChatGPT stopped being a limited demo and became a real tool for everyone.