Codex Plans and Pricing: What Each ChatGPT Tier Actually Gets You
Updated August 5, 2026
The plans at a glance
As of August 2026, Codex access maps onto the ChatGPT plans like this. There is no plan you can buy that gives you Codex and nothing else.
| Plan | Price | Codex access | Usage |
|---|---|---|---|
| Free | $0 | Limited | Enough to try it, not to work |
| Go | $8/month | Yes | The cheapest real access |
| Plus | $20/month | Yes | The 1x baseline most people mean |
| Pro | $100/month | Yes | 5x Plus |
| Pro | $200/month | Yes | 20x Plus, plus the rest of the Pro bundle |
| Business | $25/user/month ($20 annual, min. 2 users) | Yes | Team pooling and admin |
| Enterprise | Custom | Yes | Negotiated |
The $100 Pro tier is new as of April 2026 and is the single most common source of out-of-date advice about Codex pricing. Any article that says "Pro is $200" without qualification predates it. Check the current lineup before you pay, because OpenAI has moved this ladder more than once in the last year.
The two Pro tiers give you the same model suite. The difference is the multiplier, 5x against 20x, and what else comes in the box: the $200 tier bundles the rest of the Pro package while the $100 tier is deliberately the coding-focused option. If Codex is the only reason you are paying, $100 is the tier that exists for you.
Go at $8/month deserves a mention because it is genuinely the cheapest way to run a real coding agent from any of the major vendors. It is tight, but it is not a demo.
How the quota actually works: a 5 hour window plus a weekly cap
Like most of its rivals, Codex meters you on two layers. The first is a rolling 5 hour window: an intense session can exhaust it while your weekly allowance is barely touched, and the fix is to wait for the window to slide forward.
The second is a weekly cap on top. Both apply at once, so being blocked tells you nothing by itself about which ceiling you hit.
What makes Codex unusual is that the published figures are given per model, as ranges, rather than as a single number. On Plus, for example, the documented per-five-hour ranges differ substantially between the reasoning-heavy models and the fast ones, because a model that thinks longer consumes more of the same allowance. The practical consequence is that your model choice changes your effective quota more than it changes your bill, which is the opposite of how pay-per-token works.
If you keep hitting the wall, try dropping to a faster model for routine work before you upgrade a tier. On a subscription the cheaper model does not save you money, it buys you more requests inside the same window, which is usually what you actually wanted.
Credits: the pressure valve on top of the plan
When you run past your included allowance, Codex does not simply stop if you have credits on the account. Credits are a top-up balance that maps onto token usage, priced separately for input, cached input and output tokens, and they cost roughly $0.04 each as of August 2026.
This is a genuinely useful design and it changes the calculus against Claude, where hitting the weekly wall is a hard stop until the cycle resets. On Codex you can buy your way through a deadline. The flip side is that "my subscription costs $20" stops being true the moment you rely on it, so treat credits as an occasional rescue rather than part of your normal capacity.
The honest rule of thumb: if you buy credits more than once or twice a month, the next tier up is cheaper than the habit.
The pay-per-token alternative: an OpenAI API key
Codex accepts an OpenAI API key instead of a ChatGPT login, which swaps the whole quota system for per-token billing. As of August 2026, GPT-5.3-Codex is priced at $1.75 per million input tokens and $14.00 per million output tokens.
For interactive work this is usually the worse deal: coding agents are output-heavy and output is where per-token billing bites. Where it wins is automation. A CI job, a scheduled refactor, a batch of repository migrations, anything that runs unattended and must not fail because a human's personal quota happened to be exhausted, belongs on an API key with a spending cap set in the OpenAI console.
The two paths are not exclusive. Run your laptop on the subscription and give your CI pipeline its own API key, and neither can starve the other. Just make sure the key lives in the CI secret store and not in the shell profile you code in.
Is Codex free? What the free tier actually allows
Partly. The Codex CLI is a free, open-source download, and ChatGPT Free does include some Codex access. Unlike Claude Code, which is absent from its free tier entirely, you can genuinely try Codex without paying.
What you cannot do is work on it. The free allowance is sized to let you form an opinion, not to carry a project. Expect to hit the ceiling on your first real task.
The realistic budget framing: evaluating Codex costs $0, using it as a daily single-agent assistant costs $8 to $20/month, and running it as a parallel agent workflow costs $100/month. That $8 Go tier makes Codex the cheapest serious entry point in the category.
Watching Codex quota while agents run
A shared quota across three surfaces, metered on two windows, with per-model ranges rather than fixed counts, is not something you can track in your head. And the failure mode is the usual one: you find out when the agent stops.
CodeAgentSwarm, the desktop workspace for running several AI CLI agents in parallel, shows your Codex usage alongside your other agents' quotas, so you can see the wall approaching while the agents work rather than discovering it from a stalled terminal.
It pairs with the pattern from the Codex agent swarm guide: spread work across agents on different meters, so that exhausting one provider does not stop the whole session. Codex and Claude bill from entirely separate pools, which is a better reason to run both than any benchmark.
FAQ
The CLI is free and open source, and ChatGPT Free includes limited Codex access, so you can genuinely try it without paying. It is sized for evaluation rather than work: expect to hit the ceiling on your first real task. Paid access starts at $8/month on the Go plan.
As of August 2026 it depends on your ChatGPT plan: Go $8, Plus $20, Pro $100 (5x Plus) or $200 (20x Plus), Business $25 per user ($20 billed annually, minimum two users). There is no standalone Codex subscription.
The usage multiplier and the rest of the bundle, not the models. Both give the same model suite; $100 gives 5x Plus usage and is the coding-focused option, $200 gives 20x plus the full Pro package. The $100 tier launched in April 2026, which is why older articles only mention $200.
No, and this catches people out. The CLI, the web version and the IDE extension all draw from one shared allowance on your ChatGPT plan. A morning on the web reduces what your terminal can do that afternoon.
If you have credits on the account, Codex keeps working and bills against them at roughly $0.04 per credit. If you do not, it stops until the window resets. This is a meaningful difference from Claude Code, where hitting the weekly wall is a hard stop.
Subscription for interactive work, API key for automation. Per-token billing is output-heavy and coding agents produce a lot of output, but a CI job should never fail because a person exhausted their personal quota. Running both at once is a valid setup.
Codex quota is shared across the CLI, the web app and the IDE, so it drains faster than you expect when several sessions run at once. CodeAgentSwarm shows what each terminal is burning, next to every other agent you run.