Paying ChatGPT customers will soon have one budget instead of two. From December, OpenAI intends to count chat and coding usage against the same token balance for Plus, Pro and Business plans, retiring the separate pools that exist today.
The balance is denominated in credits per million tokens across all models. Under the current setup a Codex-heavy afternoon leaves chat allowances untouched. After the change, every prompt and every code generation draws from the same well.
What stays different between tiers
Plan structures do not fully converge. Pro subscriptions still avoid the five-hour limit that applies to Plus and Business, though Pro may keep weekly caps. How those caps interact with a merged balance is one of several questions OpenAI has not answered.
OpenAI has moved these limits repeatedly. Pro allowances for the 100 and 500 tiers doubled in October, and a planned reduction on the 200 tier was called off. Codex limits saw temporary doublings late last year. Teams that build workflows around a specific ceiling should assume the number moves again.
The upside and the squeeze
For developers the change removes an odd failure mode, where coding capacity runs dry while chat credits sit idle. For teams that lean hard on Codex, it introduces competition inside a single account, and the general assistant may lose out.
Business customers arguably gain the most. One metered balance is simpler to forecast than a pair of separate lines, which matters when finance teams are trying to attribute AI spend to particular projects.
OpenAI has confirmed the timing. It has not confirmed what happens to unused balance at the end of a cycle.
