OpenAI Doubles All ChatGPT Pro Usage Limits After Codex Backlash
OpenAI reversed a planned Codex quota cut and doubled every ChatGPT Pro tier overnight. Why a usage limit became as competitive as model quality.

OpenAI Doubles All ChatGPT Pro Usage Limits After Codex Backlash
One day, Codex users were posting reminders to burn through their quotas before an announced cut. The next, the cut was gone entirely — and every paid tier got doubled instead. In a post made around 1 a.m., OpenAI Codex lead Tibo announced that all ChatGPT Pro usage limits are being doubled immediately, reversing a planned reduction that had triggered a wave of developer complaints.
From halving to doubling, the turnaround took roughly a single night.
What Is Changing, and What Is Not
OpenAI says the following tiers now have doubled usage caps, effective immediately:
- ChatGPT Pro 100
- ChatGPT Pro 200
- ChatGPT Pro 500
- Business Premium
The planned reduction to the Pro 200 tier has been cancelled. Work and Codex usage allocations are unchanged. ChatGPT's Personal Agent and the dot extended limits continue, and they are still counted separately from the default quota.
Tibo's stated reason was straightforward: "We heard your kind feedback." The word "kind" did a lot of work in that sentence. The OpenAI community's reaction to the planned Codex cuts had been anything but kind, and the reversal was widely read as a direct response to that pressure.
OpenAI also repeated the usual line: better models are coming, and efficiency work continues.
From "The Reset Company" to "The Token Company"
Longtime Codex users are familiar with a single word: reset. A bug appears, the limits get reset. Users complain, the limits get reset. A promotion runs, the limits get reset. A billing error in late August triggered one reset, and another followed in October. Many developers had settled into a habit of treating quota exhaustion as a temporary state — wait a day, and it might come back.
OpenAI had, in effect, become the most closely watched AI company in the world run like a reset button.
Tibo leaned into that joke. "The Reset Company" was the old brand — the one that put out quota fires after they started. "The Token Company" is the new framing: instead of apologizing afterward, give developers tokens upfront and in larger supply. It is a shift from after-the-fact remediation to pre-emptive supply.
Why a Quota Move Moved Developers This Much
Quota changes matter because heavy Codex users are not chatting anymore. An agent sits open in the background, reading code, fixing bugs, running tests, and opening pull requests while the developer sleeps. The monthly quota is, in practice, how many hours those digital employees are allowed to work that month.
A cut to Pro 200 is the equivalent of a manager walking in and telling a team that its overnight headcount has been halved. Developers' project configurations, prompts, and daily rhythm are all arranged around a specific number, and for some people their sleep schedule is arranged around it as well. When the number moves, the whole workflow has to be re-planned.
The leverage on OpenAI is real. Claude Code is the obvious alternative, and the cost of moving between coding agents has fallen close to zero. At this point, a few extra benchmark points are not something most developers feel day to day, but running out of quota is. Quota has become as hard a competitive variable as model quality itself.
This Is Not Charity — It Is 1914 Ford
The move invites comparison with Henry Ford in 1914. Ford raised the daily factory wage from $2.34 to $5 — more than doubling it — at a moment when competitors thought he had lost his mind. Turnover collapsed, and workers on $5 a day could actually afford the product they were building: a Model T cost $490, roughly five months of wages for a worker with room and board covered.
Doubling is rarely charity. Every extra token OpenAI gives away is another weld holding a developer's workflow to Codex. The more someone uses, the harder it becomes to leave, and the compute economics eventually work themselves out.
The broader read is that, in this era, the company that gives developers more working hours is the one they follow. As for the users who spent the previous night racing their quotas — they can put that down. The limits just doubled.
More in AI News

Anthropic Leaks Claude Fable 5.5 and Next-Gen Opus Ahead of IPO
Fable 5.5 and a next-gen Opus (codename Obsidian 3) are in early testing weeks after Opus 5.5 shipped. Why Anthropic is moving this fast before its…
Claude Carter

Valve Responds to Steam's AI Game Flood: No Gatekeeping, and a Fix for the 'AI Garbage' Problem
Valve says it won't gatekeep AI-made games on Steam, admits low-quality AI titles get exposure in Next Fest, and outlines how it plans to help better…
Claude Carter

China's AI Bubble Bursts as Chip and Robotics Stocks Crash, Erasing $480 Billion
China's AI bubble is bursting. The STAR 50 has fallen 40% in under four months, chip and robotics stocks are down as much as 60%, and about $480…
Claude Carter