OpenAI Codex Pricing: Plans, Costs and Usage Limits

Superblocks Team
+2

Multiple authors

November 17, 2026

0 min read

Copied
0:00

OpenAI Codex doesn't have its own price tag. It rides along with your ChatGPT plan, from the free tier up to $200-plus Pro and per-seat Business and Enterprise, so what you pay for Codex is really what you pay for ChatGPT.

I pulled every number here straight from OpenAI's own Codex pricing page the day I wrote this, because OpenAI reshuffles these plans more often than most vendors reprice anything. Treat the figures as current-as-of-now, not carved in stone.

Most pricing guides stop at the individual plan. This one keeps going into what a team of 10, 50, or 200 engineers actually spends, and the part nobody prices at all: governing and deploying the apps your developers build once Codex hands them the code.

OpenAI Codex pricing at a glance

Plan Price (monthly) Codex access Best for
Free $0 Limited models and messages Trying Codex out
Go $8 Lightweight coding tasks Occasional, small jobs
Plus $20 Expanded usage, Codex app, plugins A few focused sessions a week
Pro From $100 (also $200, $500) Maximum tasks, no 5-hour limit Daily, heavy use
Business $20/user (annual) Shared workspace, SSO, admin controls Startups and teams
Enterprise and Edu Contact sales SCIM, RBAC, audit logs, data residency Large orgs

Verified on chatgpt.com/codex/pricing, November 2026. Business is $25/user billed monthly.

How much does OpenAI Codex cost?

OpenAI Codex pricing starts at $0 on the Free plan and scales with your ChatGPT subscription: $8 on Go, $20 on Plus, from $100 on Pro, and $20 per user on Business. There's no standalone Codex subscription to buy.

Because Codex is bundled into ChatGPT, your usage draws from a shared allowance tied to your plan, so there's no per-project or per-seat coding fee. You pay three ways: the plan itself, extra credits when you run dry, or an API key billed per token. More on each of those below.

OpenAI Codex pricing plans explained

Every paid ChatGPT tier includes Codex. What changes between them is how much you can run before the meter stops, and which models you can reach.

Free: $0 a month

The Free plan lets you test Codex on quick tasks with limited models and a small message allowance. It's enough to see what the tool does. It isn't enough to ship anything real, which is the honest con: serious work hits the ceiling fast.

Go: $8 a month

Go is the cheapest paid door in. You get a bigger allowance than Free and access to the latest models, aimed at lightweight coding tasks. The tradeoff is a modest allowance that a busy afternoon can still burn through.

Plus: $20 a month

Plus is where most individual developers start. It adds expanded Codex usage, the Codex app for macOS and Windows, and plugins that connect Slack, GitHub, and Figma, running on GPT-6.1 Sol and GPT-6 Luna. The catch: heavy daily use will push you into buying credits.

Pro: from $100 a month

Pro comes in three sizes, at $100, $200, and $500. You get the maximum Codex task allowance, and Pro plans currently carry no 5-hour limit, which is the real reason heavy users upgrade. Ultrafast mode is reserved for the $500 tier. The con is obvious: it's a big jump from $20, and you only feel the value if Codex is a daily driver.

Business: $20 per user a month

Business brings Codex into a shared workspace at $20 per user billed annually ($25 billed monthly, minimum two users). You get larger virtual machines for faster cloud runs, SAML SSO, MFA, and admin controls. What you don't get are the deepest enterprise security features, which live one tier up.

Enterprise and Edu: contact sales

Enterprise and Edu are quote-based and add SCIM, EKM, role-based access control, audit logs through the Compliance API, data residency, and priority processing. Pricing is custom and usage-scaled, so you'll talk to sales rather than read a sticker.

Three ways to pay for Codex

Codex gives you three billing routes, and picking the right one saves real money.

  • Your ChatGPT plan: the default. Usage comes out of the allowance baked into Free, Go, Plus, Pro, Business, or Enterprise. For most people this is all they ever touch.
  • Credits: when you hit your allowance, you buy more to keep going. Plus and Pro users purchase credits directly; Business, Edu, and Enterprise workspaces on flexible pricing add workspace credits.
  • The API key route: pay per token at API pricing, handy for automation in shared environments like CI.

The API route has a real cost beyond the per-token bill. You lose the cloud features: no GitHub code review, no Slack or Linear integration, no cloud chats. Model availability also follows whatever your API key can reach, not the full Codex lineup. It's great for pipelines and a poor fit for interactive building.

Codex usage limits: how the 5-hour and weekly windows work

Here's the part the Reddit threads keep arguing about. Codex doesn't sell you a flat number of messages. It meters usage against a rolling window, and the window depends on your plan.

Limits are measured as local messages per 5-hour period, and a weekly cap can apply on top. Pro plans currently have no 5-hour limit at all, which is a big part of what the higher price buys.

The ranges vary by model. OpenAI quotes them as estimates, not hard caps, because a task's real cost depends on its size:

Model Plus (per 5 hours) Standard Business
GPT-6.1 Sol 15–160 15–160
GPT-6 Astra 5–45 5–45
GPT-6 Luna 350–3,000 350–3,000

Source: ChatGPT Learn pricing docs, November 2026.

What burns the allowance fastest? Multi-step agent loops, long context, cloud tasks, image generation, and the faster speed modes (Fast runs at 2.5x, Ultrafast at 8x of the included rate). A single plain-language request can spend more than you'd guess, because the agent reads files, reasons, calls tools, and loops.

Resets run on the rolling window and aren't tied to your billing date. Blow through the allowance early, and you either wait for the reset or buy credits to keep working. GPT-6 Astra, the heaviest model, draws from your included usage allowance on the Pro tiers, while Plus and Business get more limited Astra access, with credits to extend it.

You can check where you stand anytime in the Codex usage dashboard, or run /status inside a CLI session. Treat the meter the way you'd treat a data cap: glance at it mid-week and plan intensive work around the reset.

Codex for teams and enterprise: seat costs and how to estimate spend

Once you're buying for a team, the math changes from "which plan" to "how many seats, and what happens when the heavy users run dry."

Business runs $20 per user a month billed annually, or $25 billed monthly. Enterprise and Edu are custom, with shared workspace usage, flexible or credit-based billing, and the admin controls a security review asks about first: SSO and MFA, RBAC, audit logs through the Compliance API, data residency, and no training on your data by default.

The per-seat rate makes a rough annual budget easy to sketch. At the $20 annual rate (about $240 per seat per year), here's where a few team sizes land:

  • 10 engineers: roughly $2,400 a year on Business annual billing.
  • 50 engineers: roughly $12,000 a year.
  • 200 engineers: roughly $48,000 a year, though a team that size is usually quoting Enterprise rather than paying list.

Treat those as rough floors to confirm with sales. The variable you can't model from a pricing page is credit overage: shared usage plus heavy builders means some months you'll top up, and a 200-person engineering org running Codex daily can add a real credit line on top of seats.

Other AI coding tools price seats the same way, so it's worth comparing against the way Cursor structures its enterprise tiers before you commit to one vendor across the org. Our roundup of the best AI coding assistants covers the wider field.

Is Codex cheaper than Claude Code?

The main difference between Codex and Claude Code pricing is what you get for the money, not the sticker. At matching tiers the headline prices line up:

Tier OpenAI Codex Claude Code
Entry $20 (Plus) $20 (Pro)
Mid $100 (Pro) $100 (Max)
Top $200 (Pro) $200 (Ultra)

Both bundle the coding agent into a broader subscription, and both meter usage, so neither sells you unlimited runs. So the real question is value per dollar: which one gives more usable work before you hit the wall, and which model lineup fits your stack.

That comparison deserves its own breakdown, and we did one. Our Codex vs Claude Code guide runs the tiers side by side so you can see where each lands. For a pricing decision, start there instead of trusting the matching headline numbers.

Which Codex plan should you choose?

Pick the plan that matches how hard you'll actually run it.

  • Choose Free or Go if you're kicking the tires or running the odd small script. Go's $8 is the cheapest real foothold.
  • Choose Plus ($20) if you code with Codex a few focused sessions a week. It's the sensible starting point for a working developer.
  • Choose Pro ($100–$500) if Codex is your daily driver, you run long agent sessions, and the 5-hour limit keeps interrupting you.
  • Choose Business or Enterprise if a team needs shared usage, SSO, admin controls, and an audit trail.

Is it worth it over raw API billing? For steady use, yes. A plan's included allowance works out far cheaper per token than metered API pricing, which is why most regular users stay on a subscription and only reach for an API key to handle overflow. Start low, watch the meter, and move up only when you hit the ceiling week after week.

After the code is written: governing and deploying what Codex builds

Codex pricing covers one thing: writing code. The cost nobody puts on the pricing page shows up later, once that code becomes apps running across your company.

One IT director described the pattern bluntly in a buyer survey: "We recently deployed ChatGPT and Codex across the company, and unsurprisingly, we are now running into users creating random vibe coded applications that don't conform to any of our standards." That's the real bill. Shadow AI is the new shadow IT, and the apps your engineers ship with Codex still need a governed home, access rules, and an audit trail someone can actually read.

Superblocks is the governed enterprise vibe coding platform where that happens. It coexists with Codex as the layer your team builds on once code exists, so business teams can ship safely and IT sets the guardrails once.

With Superblocks, teams can:

  • Deploy and govern apps built with Codex and other AI coding agents in one place.
  • Set guardrails once that IT controls, so non-engineers build inside safe limits.
  • Query who built what, what data it touched, and who has access through the Superblocks MCP.
  • Keep an audit log of every build, query, integration access, and package install.
  • Run the full platform inside their own AWS VPC with admin-approved models.
  • Enforce design systems and security standards automatically across every app.

Codex writes the code. Superblocks turns the result into a governed system of record and keeps it from becoming sprawl, and it covers what happens after a Codex build is ready to ship.

To see governed app building against your company's own data, read the Superblocks docs or book a demo.

Frequently asked questions

Is Codex free or paid?

Codex is free to try on the ChatGPT Free plan, but the free allowance is limited. Real work needs a paid plan, starting at $8 a month on Go or $20 on Plus.

Is Codex free with ChatGPT Plus?

Yes, Codex is included with ChatGPT Plus at $20 a month, with an expanded usage allowance over the free tier. It isn't a separate charge on top of your Plus subscription.

Is the $100 Codex plan worth it?

The $100 Pro plan is worth it for daily, heavy users, because it raises the allowance and removes the 5-hour limit that interrupts long sessions. Lighter users are better off starting on Plus and upgrading only if they keep hitting the ceiling.

Is Codex cheaper than Claude Code?

At matching tiers ($20, $100, and $200), Codex and Claude Code carry the same headline prices, so neither is cheaper on the sticker. The real difference is usage allowance and value per dollar, which our Claude Code pricing guide breaks down.

What happens when you hit your Codex usage limit?

When you hit your Codex usage limit, you either wait for the rolling reset or buy credits to keep working. Plus and Pro users purchase credits directly, Business and Enterprise add workspace credits, and anyone can fall back to an API key billed at standard API rates.

Does Codex have its own pricing plan?

No, Codex has no standalone pricing plan. Access is bundled into your ChatGPT subscription across the Free, Go, Plus, Pro, Business, and Enterprise or Edu tiers.

Stay tuned for updates

Get the latest Superblocks news and internal tooling market insights.

You've successfully signed up

Request early access

Step 1 of 2

Request early access

Step 2 of 2

You’ve been added to the waitlist!

Book a demo to skip the waitlist

Thank you for your interest!

A member of our team will be in touch soon to schedule a demo.

8

production apps built

30

days to build them

10

semi-technical builders

0

traditional developers

8+

high-impact solutions shipped

2 days

training to get builders productive

0

SQL experience required

See full story →

See the full Virgin Voyages customer story, including the apps they built and how their teams use them.

Large cruise ship sailing in a harbor with a road lined with palm trees and cars in the foreground.
Why not Replit, Lovable, or Base44?

"Those tools are great for proof of concept. But they don't connect well to existing enterprise data sources, and they don't have the governance guardrails that IT requires for production use."

Superblocks Team
+2

Multiple authors

Nov 17, 2026