ChatGPT Codex shared token budget December is the end of the all-you-can-prompt era in practice, even though the monthly price stays. Starting in December 2026, Plus, Pro and Business plans share a single token budget across ChatGPT and Codex, as OpenAI folds Work’s agentic capabilities into the main ChatGPT experience.
A token is the unit a model processes and generates. Longer, harder tasks burn more of it. That one fact explains the whole change.
What the merger actually does
Until now, chat and work lived in separate experiences with separate allowances. From December, they draw from one pool. Users see combined remaining usage and reset times in Settings, then Usage in ChatGPT, or in Usage and Billing in the Codex app.
Three rules matter:
- Instant chat does not count. Quick chats with Instant models stay outside the budget, so light use still feels flat-rate.
- Advanced models and long agent tasks do count. The heavier the model and the longer the task, the faster the pool drains.
- Hitting the limit has three exits. Wait for the reset, upgrade the plan, or buy extra credits.
OpenAI says it is preparing carefully and expects most users will not need to wait for resets more often than now. That is a reassurance, not a quota. No exact December quotas have been published, and no exact December date has been published either.
The device and card limits behind it
A separate OpenAI help page, updated around November 1, adds a harder edge. ChatGPT now limits how many paid accounts you can use on one device, and how many subscriptions one card can carry.
The limits cover Go, Plus, Pro 100, Pro 200 and Pro 500. Free, Business, Enterprise and Edu plans do not count. Logging out does not free a slot, and buying credits or resetting usage does not clear the restriction either.
Read together, the two changes point the same way: OpenAI is closing the loopholes that let heavy users stack cheap flat-rate accounts, and pointing them instead at extra credits or usage resets — which reporting notes are generally more expensive than one more subscription.
The evidence: what is confirmed and what is not
| Detail | What the source says |
|---|---|
| Change | Single token budget shared across ChatGPT and Codex |
| Plans affected | Plus, Pro and Business |
| Timing | Starting in December 2026; exact date not published |
| What does not count | Chats with Instant models |
| What counts most | Advanced models and complex, long agent tasks |
| Where to check | ChatGPT Settings, then Usage; Codex Usage and Billing |
| At the limit | Wait for reset, upgrade, or buy extra credits |
| Higher-plan preparation | OpenAI recently doubled usage limits on higher-priced plans |
| Device/card limits | Paid accounts per device and per card limited; Free, Business, Enterprise, Edu excluded |
| Unpublished | Exact per-device and per-card numbers; exact December token quotas |
Work is OpenAI’s agentic mode, where ChatGPT acts across tools and longer tasks instead of answering one chat at a time.
Budget one pool for chat and code, in 4 steps
1. Measure now — Check Settings, then Usage, before December, and note your real weekly burn.
2. Split by value — Instant models for quick questions; advanced models only where quality pays.
3. Cap agent runs — Give long Codex tasks a scope and a stop rule before they start.
4. Plan the top-up — Decide in advance whether a reset wait, an upgrade, or credits is cheapest for you.
How to profit from it
A shared meter rewards the person who can measure work. That is a service.
- Sell a usage audit. Read a client’s ChatGPT and Codex usage, separate Instant chats from agent work, and forecast the December pool.
- Rewrite agent prompts for the meter. Shorter scopes, clearer stop rules and cheaper first passes cut token burn without cutting output.
- Offer a “one pool” playbook. Small teams need one page: which tasks use which model, and what happens at the limit.
- Do not sell account stacking. Device and card limits now block it. Sell legitimate efficiency instead.
- Re-check in December. Quotas are unpublished today. A follow-up audit when they land is a second, honest engagement.
Sources
OpenAI’s updated help page as reported by The Decoder on November 1, 2026, and the November 2 daily summary at PoYo. Check OpenAI’s help center for the current wording before changing a plan.
Frequently Asked Questions
What changes in December for ChatGPT and Codex?
Plus, Pro and Business users get one shared token budget across both, as Work merges into the main ChatGPT experience.
Does every chat count against the budget?
No. Chats with Instant models do not count. Advanced models and complex agent tasks use the budget fastest.
Where can I see my remaining usage?
In ChatGPT under Settings, then Usage, or in the Codex app under Usage and Billing.
What happens when I hit the limit?
You can wait for the reset, upgrade your plan, or buy extra credits.
Are there device or card limits too?
Yes. OpenAI limits paid accounts per device and subscriptions per card for Go, Plus and the Pro tiers; logging out does not free a slot.
Has OpenAI published exact December quotas?
No. No exact quotas or exact December date have been published, so treat any specific number as a guess.
Keep Reading
Want more tools that actually pay for themselves? Browse the AI Tools section for the latest launches, pricing breakdowns, and profit playbooks — and check the official source linked above before you buy or deploy anything.