On October 29, OpenAI backed down from halving the Codex and Work quotas of its $200 Pro plan, one day before the reduction was due to take effect. In the same move, the company doubled the quotas for the Pro 100 and Pro 500 plans. The price of the plans remains unchanged, but quotas have shifted multiple times at OpenAI as well as at Anthropic.
On September 29, OpenAI wrote to some subscribers of its $200 Pro plan. Their Work and Codex quotas were set to drop from 20 to 10 times those of the Plus plan, and their weekly messages to GPT-6 Pro from 200 to 100. Existing subscribers kept the old terms until October 29 inclusive, before the company canceled the whole thing the day before the deadline.
Thibault Sottiaux, who heads Codex and ChatGPT Work at OpenAI, announced the company was abandoning this reduction following customer feedback. According to The Decoder, the Work and Codex quotas for the Pro 100 and Pro 500 plans are being doubled. The $20 Plus plan is not affected. OpenAI warns that a doubled quota does not double the number of tasks completed, as consumption varies by model, task, and settings.
Four resets between August and September at OpenAI
This reversal is part of a series. OpenAI reset the Work and Codex counters for its paying subscribers on August 8, then on August 29, then for all paid subscriptions on September 7. On September 22, the company added an “on-demand” reset, triggered by the subscriber themselves, according to the record by AI Catchup.
On August 29, Thibault Sottiaux explained that bugs in the tracking were inflating consumption and that, once fixed, the same quota would go 10 to 50% further. In early October, he promised 28 days of daily improvements or full resets. Meanwhile, on August 25, the Plus plan had its five-hour limit restored.
Anthropic has seen the same back-and-forth
On April 21, Anthropic’s pricing grid removed Claude Code from the $20 Pro plan. Less than 24 hours later, after an outcry from developers, the option was reinstated, as documented by developer Pasquale Pillitteri. On May 6, Anthropic doubled the five-hour limits for Claude Code and removed the reduction during peak hours, citing the leasing of the entire capacity of SpaceX’s Colossus 1 data center, over 220,000 GPUs, i.e., graphics processors.
A class action lawsuit filed on June 15 alleges that Anthropic delivers, on its Max plans, far less usage than the multiples promised, according to Engadget. Anthropic declined to comment to the site. Its motion to dismiss the complaint will be heard on November 6.
Why the quota moves and not the price
Anthropic’s IPO filing, made public on October 26, sheds light on the issue from Anthropic’s side. According to the S-1 filed with the SEC, subscriptions generated $3.2 billion in the first half of 2026, or 20% of revenue, with the rest coming from consumption billed per token, i.e., per unit of text processed by a model. In the third quarter, the top 5% most active Claude Enterprise users on Claude Code spent about twenty times more than those who don’t use it, the latter remaining under $75 per month.
Anthropic also acknowledged that its cloud suppliers had delivered less capacity than it requested. In its risk factors, it states it has modified, and may modify, its usage limits, rate limits, and product capabilities. OpenAI does not publish a prospectus. The Decoder links its October 29 move to competition with Anthropic, ahead of the latter’s IPO. In this model, a developer who codes every day is both the most costly subscriber on a fixed plan and an influencer for enterprise contracts.
