Your ChatGPT conversations are about to come out of the same pot as your Codex coding sessions. According to Mad Robot, OpenAI updated its help center article "Understanding ChatGPT and Codex usage limits" on Saturday to say that "starting in December, Plus, Pro, and Business plans' usage allowance will be shared across ChatGPT and Codex" as the company brings "the agentic capabilities of Work into the main ChatGPT experience." No blog post, no announcement — just a help page edit spotted on X, where one summary post passed 89,000 views in six hours.
The mechanics are simple: one allowance, measured in tokens — "the units of information a model processes and generates" — instead of separate limits for chat and coding. More advanced models and complex tasks burn more of it. Instant chat, the fast default model, is the one thing that won't draw from the pool. Your combined remaining usage and reset times will show in Settings, under Usage; when you hit the cap you wait for the reset, upgrade your plan, or buy credits.
Here's the part that should make heavy users nervous: OpenAI published no allowance sizes and no exact December start date. Nobody can tell whether the single pool will be bigger or smaller than the two separate allowances it replaces — which is exactly why developers on Hacker News are unhappy. One researcher wrote that the decoupled billing was "one big reason" they stayed a Pro customer; another said the separate limits were precisely why they preferred OpenAI over Claude, which already counts chats and Claude Code against one shared limit.
The timing is hard to read as anything but economics. The change landed days after OpenAI cancelled a planned cut to Pro 200's Work and Codex allowance and doubled Pro 100 and Pro 500 instead — and a week after it cut its latest funding target from $30 billion to $20 billion. Token metering ties each plan closer to what its requests actually cost the lab, and the model you pick now decides how far your plan goes. OpenAI says it expects "most users won't need to wait for their usage to reset more often than they do now." Treat that as a forecast, not a promise — and watch for the allowance numbers, which are the only part that actually matters.
— via Mad Robot