Last checked September 25, 2026, 2:06 a.m. Pacific. This is a living article; changes are logged below.
OpenAI’s public Codex repository now recognizes a plan identifier called promax and displays it as Pro (Max). The same change renames two other internal identifiers for display: prolite becomes Pro, while pro becomes Pro (More). That is a real, merged code change. It is not yet a public announcement of a new subscription. OpenAI’s merged Codex change
For anyone deciding whether to change plans, the missing details are the important ones: OpenAI has not published a separate Pro Max price, launch date or allowance in the pricing information checked for this article. The code also cannot tell us whether Pro (Max) will be a new product, a new name for an existing plan or part of a limited rollout.
The lead came from a September 24 post by TokenGremlin. I checked its claims against OpenAI’s merged pull request #47971 and the changed source files. Here is what they show.
The three names are in OpenAI’s code
| Plan identifier in Codex | Display name in the merged change |
|---|---|
prolite |
Pro |
pro |
Pro (More) |
promax |
Pro (Max) |
These mappings appear in Codex’s authentication code, as well as its account display, terminal status and analytics. The old labels in this file were “Pro Lite” for prolite and “Pro” for pro. The change gives all three identifiers distinct names that a person could see.
There is a trap in reading those names as a price list. The strings are software identifiers, not a subscription checkout page. No price or usage multiplier accompanies promax in the changed files.
How far does the support go?
The new identifier runs through several practical parts of Codex:
| Area | What the change establishes |
|---|---|
| Sign-in and account information | Codex can parse promax from plan data and represent it as a known account type. Authentication, account type |
| Usage and rate limits | Codex can recognize the identifier in a rate-limit response. The change supplies no Pro Max quota. |
| Client definitions | Generated TypeScript and Python types include promax. |
| Limit message | The client can direct a Pro Max user who reaches a usage limit to the page for buying more credits. That does not define a credit balance or price. Limit handling |
OpenAI added tests for plan parsing, account responses and rate-limit responses in the same pull request. The files show that OpenAI intended Codex to understand this identifier across more than one screen. They do not prove that an account can buy Pro Max today or that every production service has switched it on.
The change also adds Pro Max to a check for experimental context behavior. The surrounding function requires the right authentication mode, a supported model, enabled features and compatible provider settings. When those checks pass, it turns on a history-notes extension. There is no new Pro Max context-window size in that code. Read the full eligibility function
That is the limit of the public evidence. “Pro Max gets a bigger context window” would be a leap from an eligibility check to a benefit OpenAI has not documented.
The existing Pro plans matter here
OpenAI currently documents a $100 Pro tier with 5× the Plus usage allowance and a $200 Pro tier with 20×. On September 10, it paused new subscriptions and upgrades to the $200 tier. Existing $200 subscribers continue to renew, and a limited return option exists for some former subscribers. The $100 tier remains available. OpenAI Help Center: About ChatGPT Pro tiers
| What OpenAI documents now | Price | Usage positioning | New sign-ups |
|---|---|---|---|
| Pro $100 | $100/month | 5× Plus | Available |
| Pro $200 | $200/month | 20× Plus | Generally paused |
promax in Codex code |
Not published | Not published | No separate public offer verified |
It is tempting to put the $200 price beside “Pro (Max)” and assume they are the same thing. The available sources do not make that connection. They also do not say the $200 pause caused this code change. Those are questions for OpenAI, especially for anyone already paying for the higher tier.
Usage multipliers themselves are not guaranteed prompt counts. OpenAI says model choice, task size, context and other factors affect how quickly a plan’s allowance is used. Codex pricing documentation
Code merge versus release versus plan launch
The Pro Max pull request was merged on September 24 at 5:20 p.m. Pacific (September 25 at 00:20 UTC). That means the implementation is in the public repository’s main branch.
The authentication file in the Codex CLI 0.157.0 stable tag lacks the promax parser entry. A newer 0.158.0-alpha.13 tag includes both the parser and the three display names. That moves this change beyond the repository’s main branch into a tagged prerelease. It does not establish general availability in the stable CLI, the desktop app or OpenAI’s subscription checkout.
The public ChatGPT pricing page still listed Pro without a separate Pro Max purchase option at the latest check. We have not tested a Pro Max account or a checkout flow. The next decisive evidence will be an official offer spelling out the price, allowances, eligibility and effect on current Pro subscribers.
For now, OpenAI has made Pro (Max) a real name in Codex code. What that name will buy remains open.
Updates
- September 25, 2026, 2:06 a.m. Pacific: The Codex CLI 0.158.0-alpha.13 source tag now contains
promaxand the Pro, Pro (More) and Pro (Max) labels. Official plan pages still do not identify a separate Pro Max price, allowance or purchase option. - September 24, 2026, 8:25 p.m. Pacific: Rechecked OpenAI’s merged code, pricing documentation, Pro Help Center guidance and public pricing page. Confirmed the three display labels and the existing $200 Pro sign-up pause. No separate Pro Max price, allowance or purchase option was verified.
Trending on Kingy
Keep reading with the stories getting the most attention now.
-
GPT-6 Astra vs Claude Opus 5.5: Which Reasoning Setting Gives You the Best Value?
Read story -
GPT-6 Sol and GPT-6 Luna: Specs, Benchmarks, Pricing and How They Compare to Claude Opus 5.5, Fable 5.1 and Gemini
Read story -
Claude Opus 5.5: Specs, Benchmarks, Pricing and How It Stacks Up Against GPT-6 Astra, Fable 5.1 and Every Frontier Model
Read story
