QUICK ANSWER
The Codex Pro 2x introductory promo expired May 31, 2026. From June 1, Codex Pro ($100/month) runs at standard capacity - approximately half the effective throughput of the promo period. GitHub Copilot also moved to usage-based billing on June 1. If you have not modelled your Codex usage at standard rates, now is the time. The promo expiry does not change the plan price - it changes how much Codex work that $100/month actually buys.
What the 2x Promo Was and What It Covered
When OpenAI launched Codex Pro at $100/month, it applied an introductory 2x capacity multiplier - meaning subscribers received double the standard monthly Codex capacity for the same price. The promo was time-limited from launch and expired on May 31, 2026. OpenAI has not announced an extension.
The 2x multiplier applied to Codex task capacity - the number and length of coding tasks, computer use sessions, Goal mode runs, and parallel agent threads a Pro subscriber could run per month. From June 1, that same $100/month buys the standard Codex Pro capacity allocation. If you were using Codex Pro heavily during the promo period, your effective monthly budget for Codex work has approximately halved without any change to your billing.
The Broader Promo Expiry Wave - Three in Ten Days
The Codex Pro expiry is part of a pattern that has now repeated across every major AI tool launch since April 2026. Three introductory rates expired within ten days of each other:
| Tool |
Promo |
Expiry |
What Changes |
| Cursor Composer 2.5 |
First-week 2x rate ($0.25/$1.25 input/output) |
May 25, 2026 |
Returns to standard $0.50/$2.50 pricing |
| Codex Pro |
2x capacity multiplier on $100/month plan |
May 31, 2026 |
Standard capacity from June 1; no price change |
| GitHub Copilot |
Fixed per-seat billing |
June 1, 2026 |
Usage-based billing; no published per-credit rate at launch |
The pattern is consistent and worth naming: major AI tool launches now standardly include an introductory multiplier that makes the tool appear cheaper than it will be at steady-state. Teams that build workflows during the promo window at subsidized capacity then face 2-3x effective rate increases within 90 days. The introductory rate is not the real price - the post-promo standard rate is the real price, and that is the figure to use when evaluating whether a tool fits your budget.
GitHub Copilot's transition to usage-based billing from June 1 is the most significant of the three changes in terms of cost unpredictability. Usage-based billing means costs scale with actual usage rather than a fixed per-seat fee. For teams with highly variable Copilot usage - some developers using it constantly, others rarely - usage-based billing may reduce total cost. For power users with consistent high-volume usage, it may increase costs. OpenAI has not published a per-credit dollar rate for the new Copilot billing model as of May 31.
Which Codex Plan Is Right After June 1
ChatGPT Plus + Codex ($20/month base)
Standard Codex access with the lower usage quota. Right for: developers using Codex occasionally - a few tasks per day, no extended Goal mode sessions, no heavy parallel agent use. If you hit limits on Plus regularly, step up before defaulting to Pro.
Codex Add-On for Plus (pricing not publicly disclosed)
Additional Codex capacity on top of Plus without upgrading to Pro. Right for: Plus users who hit limits regularly but do not need the full Pro quota. Confirmed available through at least May 31, 2026 - check OpenAI's pricing page for current availability.
Codex Pro ($100/month, standard capacity from June 1)
Right for: developers running multiple parallel agents daily, large codebase refactors, extended Goal mode sessions (hours-long autonomous runs), and Windows computer use workflows. If the standard capacity rate (not the 2x promo rate) still covers your daily workload, Pro is the right tier. If you were only viable on Pro because of the 2x multiplier, model your usage at standard rates before your next billing cycle.
ChatGPT Enterprise (custom pricing)
Right for: teams of 5+ developers who need shared Codex capacity, centralized billing, and enterprise data privacy guarantees. Per-seat cost is typically lower than individual Pro plans at team scale.
How to Check Your Actual Codex Usage
Codex app version 26.527 (the Windows update that shipped May 29) added a Profile section showing usage statistics and lifetime token activity. If you are on Mac, this section was added in an earlier update. Open the Codex app, navigate to Profile, and review your usage statistics. The daily and monthly usage graphs will show you whether your typical usage pattern fits within the standard Pro capacity or was relying on the 2x promo headroom.
The token activity view is particularly useful for modelling costs: your lifetime token count divided by months of usage gives you an average monthly token consumption figure. Compare that against the published standard Pro capacity allocation - if you are consistently near or over the standard limit, staying on Pro is the right call. If you were well under the 2x limit even during peak usage months, you may have headroom to step down to Plus with the Codex add-on.
Frequently Asked Questions
Did the Codex Pro price increase on June 1?
No. The plan price remains $100/month. What changed is the effective capacity that $100 buys - the 2x introductory multiplier that doubled monthly Codex capacity expired. Same price, half the capacity headroom compared to the promo period.
What happens if I hit my Codex Pro capacity limit mid-month?
OpenAI throttles rather than hard-stops Codex usage when you approach your plan limit - you will see slower response times and queued tasks rather than an immediate cutoff. If you consistently hit the limit before month end, upgrading to Enterprise or adding team seats is the recommended path.
How does GitHub Copilot usage-based billing work from June 1?
GitHub Copilot moved from fixed per-seat pricing to usage-based billing on June 1, 2026. You pay for what you use rather than a flat monthly fee. OpenAI had not published per-credit pricing as of May 31. Check the GitHub Copilot billing documentation for the current rate card - this is the figure to use when comparing Copilot to Codex Pro on a cost-per-task basis.
Is Claude Code cheaper than Codex Pro after the promo ends?
Claude Code is priced differently - it runs on Anthropic's API with token-based pricing rather than a flat monthly plan. For developers who primarily use Codex for agentic coding tasks, comparing the cost per completed task (not per token) between Claude Code on Opus 4.8 ($5/$25 per million tokens standard) and Codex Pro at standard capacity is the right comparison. Claude Code leads on SWE-Bench complex codebase reasoning; Codex leads on terminal automation and has desktop computer use and mobile remote control built in.