OpenAI is removing rate limits on text chats for its free and Go tiers starting next week, according to reports from The Verge and TechCrunch. Free users, who currently hit caps if they chat too much, will be able to keep going indefinitely. TechCrunch adds a second detail worth noting: those same tiers are getting a “think” button for complex queries — a manual toggle for heavier reasoning.
On the surface this reads as a generosity story. It isn’t. It’s a cost story and a moat story.
Rate limits were never about fairness
Rate limits on free AI tiers exist for one reason: inference costs real money, and free users don’t pay it. When a company drops those limits, it’s signalling one of two things — either serving a text token got dramatically cheaper, or the strategic value of the usage now exceeds the marginal cost of producing it.
Both are probably true. Model efficiency has improved sharply over successive generations, and routing infrastructure means a casual “what’s the capital of Peru” query never touches the expensive model at all. Unlimited text chat is a far cheaper promise than unlimited image generation or unlimited video, which is precisely why the announcement is scoped to text. Read the fine print on any “unlimited” offer and the boundary tells you where the costs still bite.
The think button is the actual product decision
The more revealing change is the think button landing on free and Go tiers. Giving non-paying users an explicit lever for deeper reasoning does two things at once. It offloads the routing decision to the user — cheaper than having a model decide for you on every query — and it teaches people that there are two modes, one fast and one expensive.
That’s a pricing education exercise dressed as a feature. Once a free user internalises that “thinking” is the good mode, the natural next question is how much more of it they can get. The button is a demand-generation surface as much as a capability.
Land grab, not loss leader
The competitive read is straightforward. Assistant choice is sticky in a way search never quite was: people build habits, save context, and stop shopping around. In a market where Google can bundle its assistant into Android and Search by default, OpenAI’s counter is to make the standalone product cost nothing to use heavily. Removing the friction point where a free user hits a wall and considers alternatives is worth more than the tokens it saves.
The risk is what always follows a free-tier expansion — capacity strain, quiet degradation, and the possibility that “unlimited” turns out to mean unlimited-until-we-say-otherwise. Neither source describes what happens under load, and that’s the detail worth watching once the rollout actually lands.
For users, the practical advice is simple: use it. For everyone building on top of this ecosystem, the signal is that the floor of what a free AI assistant provides just moved up again — and any product whose pitch was “like ChatGPT but without the limits” needs a new pitch.