OpenAI to merge ChatGPT and Codex token allowances in December 2026

OpenAI will combine ChatGPT and Codex usage into a single token allowance for Plus, Pro, and Business subscribers starting in December 2026. The change means a heavy session with the coding agent now draws from the same balance customers use for chat, forcing paying users to budget both products against one pool.

Usage across both products will be counted against one shared credit balance, measured in credits per million tokens across all models. Tokens are the small chunks of text that AI models read and write. The move folds Codex, OpenAI's coding agent, more fully into the ChatGPT ecosystem.

How the plans differ

The unified allowance covers OpenAI's three main paid tiers, but the structures aren't identical. Pro plans don't carry a 5-hour limit, while Plus and Business plans do. Pro users may still face weekly caps.

The change follows a broader pricing overhaul. On April 2, 2026, OpenAI shifted to token-based usage pricing, replacing a per-message estimate that treated a quick question and a compute-heavy request as roughly equivalent units. The token system was built to better reflect the computational demands of different tasks, particularly for advanced models such as GPT-6.

Pro allowances have moved several times. In October 2026, limits for Pro 100 and Pro 500 doubled, and a planned reduction for Pro 200 was canceled. Codex users saw temporary doublings of their limits in late December 2025.

What changes inside your account

For developers, the shared pool removes a common annoyance: running out of Codex capacity while ChatGPT credits sit unused. It also introduces a new kind of competition inside a single account. A team that leans heavily on Codex for coding could find its general ChatGPT usage squeezed.

For Business customers, one metered balance is easier to track and forecast than two separate ones. The research findings suggest more user-friendly and financially predictable frameworks may make businesses more inclined to adopt these tools. Teams weighing how to plan around the new limits can start with ChatGPT Courses covering usage and workflow basics.

Several questions remain open. How weekly caps on Pro plans will interact with the combined allowance is one. Whether Plus and Business users will find the 5-hour limits more restrictive once Codex draws from the same balance is another.

The pricing history adds a note of caution. OpenAI has adjusted allowances multiple times, including doublings and a canceled reduction. Anyone building workflows around current limits should expect the numbers to move again.

Why this matters for support, product, and IT teams

Every prompt and every code generation will draw from the same well come December. That makes usage tracking a shared responsibility rather than a per-tool concern. Support and product teams should log which activities consume the most tokens before the switch, so they can set internal budgets that hold up when coding and chat compete for the same allowance. Leaders setting that policy can find a starting framework in AI for Executives Courses.