OpenAI Unifies ChatGPT and Codex Token Allowance for December
Paid subscribers on Plus, Pro, and Business tiers will draw from a shared token pool across conversational and coding workloads.
In short
- Starting in December 2026, OpenAI Plus, Pro, and Business plans will share a single usage allowance across ChatGPT and Codex.
- Usage will be metered in tokens rather than fixed message counts, though Instant chat remains excluded from drawing allowance.
- Users will track combined remaining balances and reset schedules in Settings under Usage, with options to wait, upgrade, or purchase credits when limits are reached.
- OpenAI has not yet published an exact December launch date or specific numerical allowance sizes for the shared tiers.
OpenAI plans to merge usage allowances for ChatGPT and Codex into a single shared balance starting in December 2026 across its Plus, Pro, and Business subscriptions. The change, detailed in an updated OpenAI Help Center article on October 31, ties consumption across both tools directly to token-based metering as the company integrates its agentic Work capabilities directly into the core ChatGPT experience.
Under the forthcoming model, subscriber usage will no longer be siloed into separate allocations for conversational generation and code development. Instead, activity across both environments will draw from a single shared credit balance, measured in tokens—the discrete units of information that models read and generate. OpenAI stated that fast conversational requests handled by Instant chat will not draw from the shared allowance, shielding baseline conversational workflows from the meter.
How the Unified Metering Model Works
Under the unified structure outlined by OpenAI, the rate at which an account expends its allowance will depend directly on task difficulty and the chosen underlying architecture. OpenAI's support documentation notes that more advanced models and computationally demanding operations will consume more of the available quota.
Account holders will be able to monitor their remaining balance and reset timers by navigating to the Usage tab inside their account settings. When a subscriber reaches the cap, OpenAI outlines three options: wait for the designated reset window, upgrade to a higher subscription tier, or purchase supplemental credits to continue accessing supported tools. OpenAI stated in its documentation that it anticipates most users will not need to wait for usage resets more frequently than under existing limits, as reported by CellCog.
On personal Plus and Pro subscriptions, shared metering is not entirely unfamiliar. OpenAI previously combined agentic usage allowances for Codex alongside ChatGPT Work, Excel, and PowerPoint integrations. The December update expands that pooled bucket to encompass advanced chat models that previously operated under independent quotas, according to FourWeekMBA.
Model Variations and Allowance Consumption
Because token consumption dictates allowance burn, higher-tier reasoning models deplete allowances substantially faster than lightweight alternatives. While OpenAI has not published explicit December figures for the shared pool, existing documentation for Work and Codex on ChatGPT Plus reveals the disparity in throughput across model families.
According to FourWeekMBA's analysis of OpenAI documentation, a standard five-hour window on a Plus plan accommodates an estimated 5 to 45 local messages using GPT-6 Astra, compared to 350 to 3,000 messages when using GPT-6 Luna. In credit-based Office feature billing cards tracked by CellCog, GPT-6 Astra output is billed at 1,250 credits per million tokens, compared to 30 credits per million tokens for GPT-5.6 Luna.
Plan pacing also retains structural variations across customer tiers. As reported by Crypto Briefing, Plus and standard Business plans operate against five-hour rate windows alongside weekly thresholds. Conversely, Pro tiers—including Pro 100, Pro 200, and Pro 500—carry no five-hour limits for Work and Codex features, though they remain subject to broader weekly quotas.
Implications for Developers and Businesses
The move marks an evolution from OpenAI's earlier message-count models. On April 2, 2026, OpenAI began transitioning product lines to token-based metrics to align consumption quotas with raw compute costs, Crypto Briefing notes.
For enterprise administrators and developers, the shared meter eliminates scenarios where users exhaust their coding allocation while leaving chat quotas idle. However, technical users who rely heavily on Codex express concern that intensive coding sessions could deplete allowances needed for daytime conversational tasks, as noted by LuminaBench.
As of November 1, 2026, OpenAI has not published an exact calendar launch date for December, nor has it revealed the total numerical token allowances allocated to Plus, Pro, or Business accounts.
Questions people ask
When does the unified ChatGPT and Codex allowance take effect?
OpenAI's help documentation states the change begins in December 2026, but the company has not published a specific calendar day.
Which OpenAI plans will share a single token allowance?
The shared allowance applies to Plus, Pro, and Business subscription plans. Free, Go, and Enterprise tiers were not cited in the update.
Does every ChatGPT prompt count against the shared allowance?
No. OpenAI confirmed that Instant chat will not draw from the shared allowance, while complex queries and advanced models will consume tokens.
Where can subscribers check their remaining allowance and reset schedules?
Subscribers can check their combined balance, reset timers, and usage data by visiting the Usage section within account settings.
Sources
- Understanding ChatGPT and Codex usage limits | OpenAI Help CenterOpenAI Help Center, official source
- ChatGPT and Codex Will Share One Usage Limit in DecemberCellCog, Nov 1, 2026
- OpenAI to merge ChatGPT and Codex token allowances in December 2026Crypto Briefing, Oct 31, 2026
- ChatGPT and Codex are getting one shared usage allowanceLuminaBench, Oct 31, 2026
- ChatGPT and Codex Will Share One Usage Allowance in DecemberFourWeekMBA, Oct 31, 2026
How this story was made: the newsroom picked it up from Hacker News, Google News and Google Search, gathered the full text of the sources above, and drafted it with AI assistance. Every factual claim was then checked against those sources before publishing (20 claims checked). Illustrations marked as AI-generated are not photographs. Spotted an error? Tell us.
Published November 1, 2026 at 07:27 UTC
