$200 ChatGPT Pro seats lose half their Codex compute on October 30

Someone who signs up for the $200 ChatGPT Pro plan this week buys from a listing that still shows 200 GPT-6 Pro messages a week, a limit that holds only until the cut lands ^[12]. OpenAI had paused new sign-ups and upgrades to the plan before bringing it back on the smaller terms ^[11].

OpenAI is pitching Pro 500 on Ultrafast and on its highest usage limits ^[3]. It aims the plan at people who regularly hit those limits, heavy Codex users among them ^[7]. What it is doing to the existing base is changing what $200 buys. A Pro user who consumes less than half of today's Work and Codex compute will not reach the new cap ^[3]. One who consumes more will run into it.

BusinessToday reports that OpenAI engineering lead Thibault Sottiaux said the $500 plan offers half the previous API cost per dollar spent ^[14]. Neither report states Pro 500's allowance as a multiple of Plus, the unit the $200 cut is written in ^[8], so whether $500 restores the old 20x level cannot be checked. The reports do support a narrower point. The $200 seat buys less from October 30, and the one published step up costs $300 a month more per seat, or $3,600 a year ^[2].

Both reports describe individual plans. Digital Trends calls Pro 500 the company's highest-usage individual tier ^[15], and OpenAI's listing refers to self-serve Plus and Pro subscribers ^[6]. A team that expenses Pro seats for its engineers takes the change one seat at a time. The budget line for a seat assumes fixed capacity at a fixed price. After October 30 the price holds and the capacity falls. Above Plus, the ladder has gone from one $200 plan in late 2024 to a $100 option added this year and now $500 ^[13][1].

Each seat sorts on two tests. One is volume: whether the person uses more than half of today's Work and Codex allowance ^[3]. The other is latency: whether waiting on GPT-6 Astra output costs them working hours.

A seat below the volume line and indifferent to speed stays at $200, and the cut does not bind. Below the line but slowed by waiting, Pro 500 is a speed purchase; OpenAI puts Ultrafast at eight times faster ^[4]. Above the line and indifferent to speed, the problem is capacity while the $500 plan leads with speed, and the one-time credit gives room to measure real usage before deciding ^[10]. A seat above the line and slowed by waiting matches the user OpenAI describes as the tier's target ^[7].