The shift arrived over the weekend without a banner. Users of ChatGPT’s free tier opened the app and found a model they had never been given before: GPT-5, OpenAI’s flagship system, had replaced the older models behind the reply box for millions of casual users. No email announced it. The change was simply there, with tighter usage limits than subscribers face but the same underlying intelligence underneath.
OpenAI has begun deploying GPT-5 to free and Plus users, the company said, the first time the model has been made available to the general public since it was unveiled. Free accounts receive limited access, with caps that reset on a rolling schedule. Plus subscribers, who pay $20 a month, get higher rate limits, faster responses and advanced features. TechCrunch reported the expansion, describing it as GPT-5’s first broad public release.
The rollout has been incremental by design. OpenAI shipped GPT-5 first to narrower groups after its debut, letting enterprise teams and business accounts run it before the consumer base was ready. This weekend’s expansion closes that gap. The company said access would reach users across the web, mobile apps and desktop over the coming days rather than in a single switch.
What the two tiers get
The gap between free and paid is about volume, not capability. Free users can run GPT-5 for a limited number of messages before the app switches them to a smaller, faster variant built for routine queries. Plus subscribers can keep GPT-5 as their default through the day, with allowances several times larger, according to OpenAI’s plan descriptions. When a free user exhausts the ceiling, the app shows a reset time, and often a payment screen.
Features differ alongside the limits. Plus accounts unlock deeper reasoning modes, image generation with extended thinking, larger file uploads, expanded memory and access to agentic tools that can act across apps and files. OpenAI describes the tier as advanced intelligence for people who use the model for work: writing, coding, research and analysis. Free users get the core model, a simplified picker and none of the extras.
The rollout also changes how users steer the model. Paid subscribers can force GPT-5’s thinking mode, making it reason longer before answering. Free users see fewer switches. OpenAI has cautioned that full reasoning on the free tier may take several days to reach everyone as the company scales capacity; the opening is a logistics exercise as much as a product decision.
A model built for everyone
GPT-5 departs from the models that came before it. Rather than a single, one-speed system, it decides when to answer quickly and when to think, routing simple questions to instant replies and harder ones to extended reasoning. OpenAI has said the design was meant to make advanced intelligence the default setting rather than a premium toggle. In the company’s own evaluations, external experts preferred the pro-tier version’s answers on more than two-thirds of more than 1,000 real-world reasoning prompts.
The stakes reach beyond consumer chat. The release drops GPT-5 into a three-way race with Anthropic’s Claude and Google’s Gemini, both of which have pushed aggressive free tiers of their own. Google distributes Gemini through search and Workspace, reaching billions of users without a subscription. Anthropic has concentrated on enterprise contracts and developer trust. OpenAI’s counter is distribution: put the best model in front of the widest audience, then sell them the headroom.
The model also matters to developers. GPT-5 sits at the core of OpenAI’s API, and the reasoning system is what powers its agentic products, including Codex for coding. A consumer rollout that normalizes the model at scale feeds directly into that business: more people who learn to rely on GPT-5 in chat become the developers and managers who request it at work, where seats cost real money.
The business logic of a free flagship
Giving away the flagship looks expensive; the math favors scale. OpenAI’s revenue rests on two engines: subscriptions and API usage from developers. A model that millions more people can try widens the top of that funnel, and usage caps do the conversion work, pushing a user who hits the ceiling mid-project toward a payment screen.
The move also answers pressure on OpenAI’s growth. Analysts have pointed this year to slowing momentum in both consumer and enterprise segments, and the company has restructured its product line around agents and new models. Opening GPT-5 to everyone is the most direct defense: put the best model in front of the most people, then charge for the room to use it.
Costs follow. Every free request runs on inference infrastructure that OpenAI pays for, and the company has throttled capacity in past launches when demand spiked. Free-tier limits are dynamic, adjusted for demand and system conditions, which gives OpenAI room to tighten the spigot without announcing a policy change.
Bottom line
GPT-5’s arrival on the free tier resets the terms of the consumer AI market. The best model a company ships is now the default for anyone with a phone and an internet connection. OpenAI is betting that generosity at the entry point converts into subscriptions at scale, and that distribution beats exclusivity in a market where rivals give their best work away. The pace of the rollout, and the patience of free users meeting their caps, will show whether the bet holds.
