Overview
- OpenAI announced that starting in December 2026 paid subscribers on Plus, Pro, and Business will draw ChatGPT and Codex usage from a single token‑based allowance rather than separate buckets.
- Usage will be measured in credits per million tokens, the same token pricing OpenAI introduced in April 2026 that charges heavier, more complex requests more credits than simple prompts.
- Instant chat is explicitly excluded from the metered pool so quick, lightweight conversations will not reduce the shared token balance.
- Key operational details remain undisclosed: OpenAI has not published an exact December launch date, the size of each plan’s combined pool, or how existing Pro weekly caps and Plus/Business hour limits will interact with the shared allowance.
- For users this removes stranded credits between products but creates direct trade‑offs—heavy Codex coding runs can deplete the pool and reduce ChatGPT capacity—so subscribers should monitor Settings → Usage and expect numbers to change as OpenAI has adjusted allowances before.