ChatGPT Free Users Get Unlimited Text Chats with GPT-5.6 Luna
π Table of Contents
OpenAI has quietly dismantled what was arguably the sharpest edge of its paywall. As of this week, all ChatGPT users β including those on the free tier β get unlimited text-based chats, with the rate limits that once metered casual conversation simply gone. The change, announced on August 6 and rolled out starting the week of August 10, lands alongside a model shake-up: GPT-5.6 Luna becomes the default for Free and Go users, and Plus and Pro subscribers get a rewritten GPT-5.6 Sol tuned for everyday questions. OpenAI says ChatGPT has now crossed 1 billion weekly users β a milestone the company announced in the same update, and one that gives the unlimited shift its real weight.
The paywall's sharpest edge just vanished
For most of ChatGPT's life, the free tier's message counter was the most visible reason to upgrade. Hit a ceiling mid-conversation, wait for the meter to reset, or pay up β that friction sold a lot of subscriptions. By removing it entirely, OpenAI is betting that generosity now buys something more valuable than subscriptions later: default status.
It's a deliberate sequence. The updated Sol shipped to Plus and Pro immediately on August 6, Luna and the new Think button followed that week, and the unlimited text rollout completed the picture β "next week," OpenAI said at the announcement, with TechCrunch and Help Net Security confirming the staggered timeline. The GPT-5.6 family β Sol, Terra, and Luna β debuted in July, so this is an expansion of an existing lineup, not a new model generation.
What's actually unlimited
The word that matters here is "text." OpenAI is removing limits on text-based chats β not on everything ChatGPT can do:
- Unlimited text chats for every tier, free included
- Still capped: file uploads, image understanding, voice conversations, and image generation keep their separate limits
- Think button (Free and Go): a one-tap mode for higher-reasoning answers on harder questions
The distinction matters for power users who live in the file-upload and voice flows β their meters haven't moved. But for the majority who just want to ask questions, argue with the bot, and draft without watching a counter, the change is effectively total. As almcorp's coverage puts it, the free tier now behaves like a subscription for the most common activity on the platform.
Unlimited text chat turns ChatGPT from a metered tool into a default habit β and habits are the hardest thing for a competitor to dislodge.
Luna takes the wheel for free users
The free experience also gets a smarter default. GPT-5.6 Luna replaces GPT-5.5 for Free and Go users, bringing the 5.6 generation's improvements to the tier where most of those 1 billion weekly users actually live. The Think button gives that audience a taste of the deeper-reasoning behavior that was once a premium feature.
On quality, OpenAI reports that internal evaluations show factual errors down 62% for Luna compared with GPT-5.5-Instant. That figure is company-reported and has not been independently verified β treat it as a directional claim, not an audited result.
Plus and Pro get a sharper Sol
Paid users aren't just watching the free tier improve. The updated GPT-5.6 Sol is a different flavor of the same model: tuned for shorter, more direct answers on everyday tasks, rather than the long-horizon agentic work handled by the separate Codex/Work Sol, which is unchanged. A new thinking slider lets Plus and Pro users dial reasoning depth up or down β useful for trading latency against complexity on routine questions.
OpenAI cites a 68% reduction in factual errors for Sol versus GPT-5.5-Instant on its internal evals β again, an unverified vendor claim. The pattern across both models is consistent, though: the 5.6 generation is being positioned as both faster and more reliable in everyday use, which is precisely what an unlimited free tier needs to be.
Free-tier generosity becomes the battleground
The competitive logic here is hard to miss. Google, Meta, and xAI all compete for the same consumer attention, and the frontier of that fight has moved from model-vs-model benchmarks to who can afford to give the most away. By making unlimited text free while crossing a billion weekly users, OpenAI is signaling that it intends to win on defaults β the apps people open first, the assistants they trust with their daily questions.
It's a high-stakes wager. Unlimited usage at scale means real inference costs, and the bet only pays off if the free tier converts into Plus subscriptions, API revenue, and habit. But for the user, the direction of travel is unambiguous: the paywall's sharpest edge β the message counter β is gone, and the industry's next moves will be measured against that.