Codex Pricing in 2026: What Free, Plus, Pro, Business, and API Really Cost

Since October 29, 2026, OpenAI's Pro 100 and Pro 500 plans include twice their previous Work and Codex allowance, and every Pro 200 subscription has the higher allowance after OpenAI cancelled the planned cut. This helps developers whose Pro allowance interrupts paid work; the change applies automatically to existing subscriptions. Plus remains $20/month, was not part of that October allowance increase, and is still the right starting point for most solo builders. October 29 announcement, current Pro terms.

OpenAI Codex is not a separate subscription. It is bundled into ChatGPT plans, then gated by usage limits, credits, or raw API token billing depending on how you access it.

The trap is reading the monthly price as the whole cost. Codex pricing has two layers: the plan you buy and the amount of agent work your tasks consume. A file-level bug fix and a repo-wide refactor do not spend the same allowance, even if both look like a single message.

At a glance

These monthly USD prices, model access, and usage terms were verified on November 2, 2026 against ChatGPT pricing and OpenAI's Codex pricing page; the Pro tiers help article gives the current allowance differences, and Business pricing lists Standard and Premium seats.

The Terra plan lists differ across OpenAI's pages. The launch announcement names Plus, Pro, Enterprise and Edu; the October 26 changelog and Codex pricing guide also name Business. ChatGPT's general pricing comparison has no GPT-6.1 Terra row yet and still lists limited GPT-5.6 Terra desktop access for Free and Go. The paid-plan entries above follow the Codex-specific guidance; model visibility still depends on the client and workspace.

The short answer: start on Plus, not Pro

Verified November 2, 2026: OpenAI's Codex pricing page still lists Plus at $20/month and Pro at $100, $200, or $500/month.

Since October 26, 2026, GPT-6.1 Terra gives Plus users another way to stretch the same Codex allowance. OpenAI says it covers more tasks than GPT-5.6 Terra and substantially more than GPT-6.1 Sol or GPT-6 Astra, without changing the overall plan limits at that launch. Try it for suitable everyday work before upgrading; keep Sol or Astra for the most complex coding. Launch announcement, current usage terms.

Since October 14, 2026, GPT-5.5 has been retired from Codex with ChatGPT sign-in across all plans. Replace saved gpt-5.5 selections with GPT-6.1 Sol on Plus, Pro, Business, Enterprise or Edu when available, or GPT-6 Luna on Free and Go in the desktop app when available. The OpenAI API and Codex sessions authenticated with an API key are unaffected.

Plus is the first serious Codex plan because it includes the surfaces builders actually use: Codex on the web, the CLI (command-line interface), the IDE extension in your code editor, and iOS. It also includes cloud-based integrations such as automatic code review and Slack integration. Its current Codex model lineup includes GPT-6.1 Terra, GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol and GPT-6 Luna. Astra access here is separate from GPT-6 Pro in Chat; Plus includes Astra in Work and Codex.

That is enough for a founder fixing a bug, an operator changing an internal tool, or a solo developer giving Codex scoped tasks between normal coding sessions. The $20 plan gets expensive only when it creates stop-start work. If you exhaust your allowance between tasks and spend an hour waiting for a reset, the real cost is not $20. It is the lost session.

Pro starts at $100/month, with $200 and $500 options. Pro 200 includes more usage than Pro 100; Pro 500 has the highest included Pro allowance. All three currently have no five-hour Work or Codex limit, but they still have an included usage allowance. Buy the upgrade when Plus still cuts off work you need to finish after matching the model to the task, and check the allowance and terms shown to your account before paying more.

If you are comparing Codex with Claude Code, use the dedicated Codex vs Claude Code breakdown. This page is the Codex pricing decision by itself: which plan to buy, what the limits mean, and when token billing matters.

Codex pricing: what each plan buys

OpenAI's Codex pricing page splits Codex into three buying modes: personal ChatGPT plans, workspace plans, and API-key usage. The personal ladder is Free at $0, Go at $8, Plus at $20, and Pro at $100, $200, or $500 per month.

Free and Go are useful for proving whether Codex fits your workflow. Go at $8 is cheap, but its positioning around lightweight coding tasks is the phrase that matters. The Codex pricing guide lists GPT-6 Luna at Standard speed in the desktop app for both, subject to rollout. ChatGPT's pricing table also still lists limited GPT-5.6 Terra access in Work and Codex on desktop for Free and Go. Neither page gives these plans a fixed Codex task count. Treat them as a low-friction test before choosing a daily coding budget.

Plus at $20/month is the first paid plan that maps to normal solo work. It gives you the local and web surfaces, plus the integration layer that makes Codex more than a terminal command. If your work is scoped, such as "add a form validation test", "trace this regression", or "update this component to match the new API", Plus is the right first buy.

All Pro tiers keep the Plus features and current model lineup. Pro currently has no five-hour Work or Codex limit. Pro 100 costs $100/month; Pro 200 costs $200/month with more included usage; Pro 500 costs $500/month with the highest included usage and Ultrafast. New and existing Pro 200 subscriptions now receive the same higher Work and Codex allowance. The upgrade decision below uses those published terms.

Business Standard and Premium list the same current model lineup: GPT-6.1 Terra, GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol and GPT-6 Luna, subject to workspace and client availability. Business brings workspace controls, SAML SSO (single sign-on), MFA (multi-factor authentication), larger virtual machines for cloud chats, optional credits, and no training on business data by default. Business Standard is $20/user/month for at least two users when billed annually or $25/user/month when billed monthly, with the same local-message estimates as Plus. Business Premium is $100/user/month annually or $125/user/month monthly; OpenAI now states more usage than Standard with no five-hour limit, without a numerical multiple.

Enterprise and Edu use contact-sales pricing. They add priority processing, enterprise security and controls, audit logs, and data retention and residency controls. Flexible pricing has no fixed rate limits and scales with credits; without flexible pricing, most features have the same per-seat limits as Plus. GPT-6.1 Sol is off by default in Enterprise and Edu until an administrator enables it, and access to the respective Work or Codex surface is also required. Workspace model controls also govern GPT-6 Astra, GPT-6 Sol and GPT-6 Luna; changing a saved model name does not grant access.

API-key usage is a different product shape. It gives you Codex in the CLI, SDK (software development kit), or IDE extension and bills by tokens. It does not include cloud-based features such as GitHub code review or Slack integration, and model availability follows the models available to your API key.

What changes in December for Codex-heavy users

Starting in December, Plus, Pro and Business will share one usage allowance across ChatGPT and Codex, measured in tokens. Instant chat will not draw from it. Verified November 2, 2026 against OpenAI's shared-usage announcement: it gives no exact start day or token allotment per plan, and announces no change for Free, Go or Enterprise.

For a developer or team lead, the budgeting change is one pool instead of separate ChatGPT and Codex limits. Heavy non-Instant chat work can leave less allowance for coding. Tokens measure information processed and generated, including prompts, files, conversation history, tool results and responses; model choice and task complexity affect consumption. How OpenAI counts tokens.

Watch Settings > Usage for the combined remaining allowance and reset times. Before a substantial coding task, check what remains after chat work; at the limit, review the options to wait, upgrade or buy credits for eligible features. The current message estimates below are not a promised December token budget.

The five-hour window: Plus and Standard Business

As of November 2, 2026, the five-hour window shapes Plus and Standard Business under the currently published Codex terms. Pro plans currently have no five-hour limit in Work and Codex, but they still have an included allowance. An old Pro message count per window no longer describes the buying decision.

OpenAI lists usage ranges because a "message" is not a fixed unit. A small script may consume a fraction of the allowance. A larger codebase, a long-running task, or a session that needs more context can consume much more. Currently, Codex and ChatGPT Work share usage; eligible Word, Excel and PowerPoint usage also draws from the shared agentic allowance on Plus and Pro.

ChatGPT pricing describes limited Codex access on Free, expanded usage on Plus, and longer sessions on Pro without a fixed task count. The Codex pricing estimates below are the same for Plus and Standard Business, per five hours:

• GPT-6 Astra: 5-45 local messages.
• GPT-6.1 Sol: 15-160 local messages.
• GPT-6.1 Terra: 35-285 local messages.
• GPT-6 Sol: 15-150 local messages.
• GPT-6 Luna: 350-3,000 local messages.

These are estimates, not fixed message limits or guaranteed task counts. Terra's range is published by OpenAI, not calculated from its API price. The older GPT-6 Sol row still appears alongside GPT-6.1 Sol on the pricing page. Check the usage dashboard for your current allowance and reset times.

The wide ranges are the point. Codex is not billing your intent. It is consuming compute. A clean task in a small repo may fit near the high end of the range. A vague instruction in a large repo can land near the low end because Codex reads more files, carries more context, and loops longer.

Think of the window like a workbench reservation. Plus is aimed at a few focused sessions each week. Pro removes the five-hour scheduling constraint, but its total allowance still needs to be checked against your workload.

Local messages and cloud chats share your plan's allowance, and weekly limits may also apply. Cloud tasks can consume more allowance than local messages. The usage view below is the place to check which limit applies before starting another task.

Credits are the billing units used to extend eligible usage. The Standard credit rates, per 1M tokens in input / cached input / output order, are:

• GPT-6.1 Terra: 25 / 1.25 / 125 credits.
• GPT-6.1 Sol: 50 / 2.5 / 250 credits.
• GPT-6 Astra: 250 / 25 / 1,250 credits.
• GPT-6 Luna: 2.5 / 0.25 / 12.5 credits.
• GPT-6 Sol: 50 / 5 / 250 credits.

Cached input is context the model can reuse. Codex credit billing has no separate cache-write charge. Included usage is used first; after you reach its limits, eligible activity draws from your available credit balance. Buying credits does not increase your included allowance or change your plan, and credits do not cover every ChatGPT model or feature. Credit purchase prices depend on your plan or agreement, and these rates do not determine included task counts. Use the usage dashboard rather than converting API dollars into subscription messages.

Codex usage: remaining allowance and reset times

Check the Codex usage dashboard instructions or Settings > Usage in ChatGPT for your remaining allowance and reset times. In an active Codex CLI session, /status shows remaining limits. OpenAI also names Usage & Billing in Codex for your credit balance and purchase options. Usage documentation, credit controls.

On iPhone, the October 2, 2026 ChatGPT iOS release added Home Screen and Lock Screen widgets that show remaining usage and reset times.

Where five-hour and weekly limits both apply, you need allowance left in both to continue on included usage. Reaching either limit means checking its reset time, then waiting or using an eligible credit or reset option shown to your account. Switching models does not refill a shared allowance. A new five-hour window starts with your first Work or Codex message after the previous window ends. Pro currently has no five-hour limit, but that does not make its included usage unlimited. OpenAI's usage guide.

Under the current pre-December rules, GPT-6 Pro's weekly message limit in Chat is separate from the shared Work and Codex allowance. It is unchanged on Pro 100 and Pro 500; on Pro 200, accounts with the lower Chat limit now match those with the higher limit. Current Pro limits.

If a limit is reached during an active turn, Codex can continue that turn subject to fair-use limits. Plus and Pro users can buy credits; Business, Enterprise and Edu with flexible pricing can buy workspace credits. Extra local chats can also run on a separately billed API key. What happens at the limit.

When Pro is worth the upgrade

Pro is worth considering when Codex interrupts paid work even after you use Terra for suitable tasks and reserve Sol or Astra for harder coding. Start by comparing Plus with the $100 plan; choose $200 or $500 only when the terms available to your account justify the extra spend.

All three Pro plans currently have no five-hour Work or Codex limit, but each still has an included allowance. Their buying decisions differ:

• Pro 100 ($100/month): the least expensive Pro plan, now with twice its previous Work and Codex allowance. It still includes less usage than Pro 200 and no Ultrafast. Consider it first when Plus regularly cuts off a working session despite appropriate model choices.
• Pro 200 ($200/month): more included usage than Pro 100, with the higher allowance now available to every subscriber. The planned reduction is cancelled; this tier does not include Ultrafast.
• Pro 500 ($500/month): twice its previous Work and Codex allowance, still the highest included usage of the Pro plans and the only Pro tier with Ultrafast. Buy it when that access or additional allowance matters to your work.

GPT-6.1 Sol Ultrafast: who gets it and what it spends

Since October 8, 2026, GPT-6.1 Sol Ultrafast has been available in Codex and ChatGPT Work on Pro $500 and eligible Enterprise and Edu plans, alongside GPT-6 Astra Ultrafast. Pro $500 spends its included allowance first, then available credits. For either model, the pricing page lists 8x Standard consumption for included usage and 6x Standard rates for purchased credits and Enterprise pay-as-you-go usage. These are usage and billing multipliers, not speed guarantees. Release date, pricing and rates.

Eligible Enterprise agreements use credits or USD usage-based billing; eligible Edu plans use credits. Legacy Enterprise plans based on rate limits do not qualify. Enterprise Ultrafast is off by default until a workspace owner enables it for selected users or the workspace; existing per-user spend controls apply. Other self-serve plans, including Business, do not gain Ultrafast by buying credits. Plan and workspace requirements.

For a solo builder, this gives $500 a specific reason to exist, but does not change the $20 starting recommendation. Pay for Ultrafast when waiting on the model is costly enough to justify both the plan price and faster allowance consumption.

Pro 200 allowance: the planned cut is cancelled

Pro 200 now has one Work and Codex allowance for all subscribers. Accounts on the lower allowance received twice their previous allowance to match the higher level; accounts already at that level keep it. New and returning subscribers receive the same allowance, at the unchanged $200/month price. Active subscribers do not need to cancel, resubscribe, or wait for renewal. Buying credits on Pro 100 or Pro 200 still does not enable Ultrafast.

OpenAI publishes no exact Pro allowance totals or fixed per-tier task counts. The increase is not a promise of twice as many completed tasks: model choice, context, reasoning, tools and speed settings still determine consumption. Use the account's usage display to judge whether the new allowance carries your workload.

For a solo technical founder, Plus is the default. If you are building a feature after dinner, fixing bugs between calls, and asking Codex for bounded changes, the $20 plan is hard to beat.

For a senior engineer using Codex during implementation blocks, the first Pro comparison is $100 versus $20. That is an extra $80/month. Judge it by interruptions you can avoid and the allowance your account shows, rather than an assumed usage multiple.

For an agency owner or internal-tools builder running Codex across several repos, Pro 200 buys more included usage than Pro 100, with the higher allowance now available to all subscribers. Recheck whether that allowance carries your work before budgeting for a more expensive tier. Choose Pro 500 when Ultrafast is a specific requirement. Several active repos do not, by themselves, establish which Pro tier you need.

If you are deciding between OpenAI and Anthropic plans, pair this with the Claude Code pricing breakdown. Compare how often each agent stops your work, what features you lose by changing tools, and which model is better for your codebase.

When API pricing matters

Use OpenAI's API pricing page when Codex is part of automation instead of a human ChatGPT session. A developer using the CLI with a ChatGPT login is usually choosing a plan. A team wiring Codex into CI (continuous integration), a custom SDK workflow, or shared automation should look at token billing.

The listed rates apply to Standard processing with up to 272K input tokens. Cache writes are a separate API rate: with 30-minute retention, $1.25 per 1M tokens for GPT-6.1 Terra, $2.50 for GPT-6.1 Sol, $12.50 for GPT-6 Astra, and $0.125 for GPT-6 Luna. The pricing page lists separate rates for 12-hour retention. The model pages for Terra, Sol, Astra and Luna specify that above 272K input tokens, the full request is billed at 2x input and cache rates and 1.5x output rates. API billing is separate from ChatGPT subscriptions. GPT-6.1 Terra is positioned for visual reasoning, computer use and writing, Sol for complex coding and agentic workflows, Astra for the hardest end-to-end work, and Luna for focused, high-volume tasks.

OpenAI describes Terra's Standard API pricing as 50% below GPT-5.6 Terra and GPT-6.1 Sol. That is an API price comparison, not a promise of a fixed multiple of subscription tasks. Terra launch pricing.

The API path bills what you use, which fits a CI pipeline that runs only on pull requests. A solo builder who wants a predictable monthly fee and the ChatGPT cloud features should start with a subscription. API token prices do not predict how many tasks a subscription includes.

API-key Codex also has a product tradeoff. It works in the CLI, SDK, or IDE extension, but it does not include cloud-based features such as GitHub code review or Slack integration. If those features are why you want Codex, stay inside the ChatGPT plan path.

How to make your Codex limit last longer

Most Codex limit problems start as scope problems. Before upgrading, reduce the amount of context Codex has to carry.

1. 1

   Give Codex a smaller job

   Ask for a concrete change at a time: "add tests for this function", "fix this failing route", or "explain why this migration fails". Broad instructions make Codex inspect more files and spend more of the window.

2. 2

   Try Terra before upgrading from Plus

   If Sol regularly runs down your allowance, try GPT-6.1 Terra for suitable everyday tasks and compare the result with what you need. OpenAI lists Terra at 35-285 estimated local messages per five hours on Plus and Standard Business, versus 15-160 for Sol. Keep Sol or Astra for the most complex coding; Luna remains an option for focused, repeatable tasks, with 350-3,000 estimated local messages. These are published ranges, not fixed task ratios. Usage estimates.

   In the desktop app, open Advanced and select Terra. The models page requires Codex CLI 0.159.0 or later: start with codex -m gpt-6.1-terra, or use /model in an interactive session. Choose the cheaper model while allowance remains; it cannot restore usage already spent.

3. 3

   Trim AGENTS.md and prompt context

   Long project instructions help only when they are relevant. If a repo-level AGENTS.md injects a wall of process into every task, split guidance into nested files so small tasks do not carry the whole company handbook.

4. 4

   Disable unused MCP servers

   MCP servers connect Codex to tools and can add context. Keep the servers you need for the task and turn off the ones that are just available by habit.

5. 5

   Avoid faster modes unless turnaround justifies the usage

   Where available, Fast mode consumes included allowance at 2.5x Standard for the same model; Ultrafast for GPT-6 Astra or GPT-6.1 Sol uses 8x. Purchased credits and Enterprise pay-as-you-go usage use 2x Standard rates for Fast and 6x for Ultrafast. These are consumption rates, not performance multiples. Start with Standard when your budget matters more than turnaround.

6. 6

   Do image work outside Codex when coding allowance matters

   Image generations use included limits 3-5x faster on average than similar turns without image generation. If the goal is code, keep visual generation out of the Codex session.

These steps do not make Plus behave like Pro. They make sure you upgrade for real demand rather than messy prompts.

Who should pick which Codex plan

Pick the cheapest plan that does not interrupt the job you actually do. Codex is valuable when it shortens the path from intention to working code. It is annoying when the plan becomes a scheduling constraint.

Use Free for a quick trial of the workflow or Go at $8 for lightweight fixes. A solo builder doing scoped work should start with Plus at $20. If Plus regularly runs out while you use Sol, try Terra for suitable work within the same allowance before comparing Pro 100. Upgrade if the work still needs more usage, or if it needs Sol or Astra and cannot tolerate the interruptions.

Pro 200 includes more usage than Pro 100, and new subscriptions now receive the same higher allowance as existing subscribers. Choose Pro 500 for the highest included Pro usage or when Ultrafast access is required. Compare the allowance shown in your account with your workload; the monthly price alone does not establish a fixed task count.

A startup with shared repos and governance needs should consider Business. A larger company needing audit and data controls should contact sales for Enterprise or Edu. For CI and custom automation, use API-key token billing.

The blunt rule: Plus is the default, Terra is the first model to try when Sol spends your allowance too quickly, $100 is the first Pro comparison, $200 needs a confirmed reason to spend more, $500 is the Ultrafast option, and API is for systems.

The plans to avoid

Free is the plan to avoid for real work. It is useful for a quick look, but it is not a dependable build setup.

Go is easy to overestimate. At $8 it looks like a bargain, but it is positioned for lightweight coding tasks. If Codex is going to touch real project code every week, Plus is the cleaner baseline.

Business is the wrong buy for a single independent builder. Its value is workspace control: admin controls, SSO, MFA, optional credits, no training on business data by default, and larger virtual machines for cloud tasks. If you do not need those, buy a personal plan.

API-key usage is the wrong buy when you want the ChatGPT Codex product experience. It is powerful, but it is not a replacement for cloud code review, Slack, and the bundled ChatGPT plan surfaces.

If none of the Codex plans fit, the next question is not "which Codex tier is hidden?" It is whether another coding agent matches your workflow better. The Claude Code alternatives guide is the better next read when the blocker is fit rather than price.

How much does Codex cost?

Codex is bundled into ChatGPT plans: Free is $0/month, Go is $8/month, Plus is $20/month, Pro 100, Pro 200 and Pro 500 cost $100, $200 and $500/month respectively, and Business Standard is $20/user/month annually for at least two users or $25 monthly. Business Premium is $100/user/month annually or $125 monthly. Enterprise and Edu use contact-sales pricing; API-key usage bills by tokens. Pro plans currently have no five-hour Work or Codex limit. Pro 100 and Pro 500 have increased allowances, and all Pro 200 subscribers now receive the higher allowance with the planned cut cancelled. Pro 200 includes more usage than Pro 100; Pro 500 has the highest included Pro usage and is the only Pro tier with Ultrafast.

Is Codex free or paid?

Both. Free includes GPT-6 Luna at Standard speed in the desktop app, subject to rollout. Daily coding with scoped tasks is better started on Plus at $20/month for web, CLI, IDE extension, and iOS access, then judged against your actual usage.

Is Codex free with ChatGPT Plus?

Yes. ChatGPT Plus includes Codex at $20/month with access across web, CLI, IDE extension, and iOS. Its current Codex models include GPT-6.1 Terra, GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol and GPT-6 Luna, subject to client availability. GPT-5.5 is no longer included with ChatGPT sign-in.

Is Codex free with ChatGPT Go?

Go costs $8/month and includes GPT-6 Luna at Standard speed in the desktop app, subject to rollout, for lightweight coding tasks. It is cheaper than Plus, but Plus is the better baseline if Codex will touch real project work every week.

Is Codex really worth it?

Codex is worth it when the value of saved time exceeds the plan fee. Compare the time saved on debugging, tests, and scoped refactors with the $20 Plus cost. If Sol runs through your Plus allowance, try Terra for suitable tasks first. If usage still interrupts paid work most days, compare the $100 Pro plan and its actual allowance with the extra $80/month it costs.

How much does Codex API usage cost?

For Standard processing with up to 272K input tokens, GPT-6.1 Terra is $1 input, $0.05 cached input, and $5 output per 1M tokens; GPT-6.1 Sol is $2, $0.10, and $10; GPT-6 Astra is $10, $1, and $50; and GPT-6 Luna is $0.10, $0.01, and $0.50. Cache writes are billed separately: with 30-minute retention, $1.25, $2.50, $12.50, and $0.125 per 1M tokens respectively. Longer retention has separate rates. API usage is billed separately from ChatGPT subscriptions.

Where can I check my remaining Codex usage?

Use the Codex usage dashboard instructions or Settings > Usage in ChatGPT for remaining allowance and reset times, and /status in an active Codex CLI session for remaining limits. ChatGPT's iOS Home Screen and Lock Screen usage widgets also show remaining usage and reset times.

Which Codex plans include Ultrafast?

Pro $500 and eligible Enterprise and Edu plans include Ultrafast for GPT-6 Astra and GPT-6.1 Sol. Enterprise owners must enable it. Other self-serve plans do not get it by purchasing credits. The pricing page lists 8x Standard consumption for included allowance and 6x Standard rates for purchased credits or Enterprise pay-as-you-go usage.

What happens when I hit a Codex usage limit?

Currently, Work and Codex share usage. Starting in December, Plus, Pro and Business share a token-measured allowance across ChatGPT and Codex; Instant chat does not consume it. Settings > Usage will show the combined balance and reset times.

Where five-hour and weekly limits both apply, you need allowance remaining in both for included usage. Pro currently has no five-hour Work or Codex limit, but its included allowance can still run out. Check Settings > Usage for the affected window and reset time, then wait or use an eligible credit or reset option. Included usage is spent first; eligible activity then draws from your available credits. An active turn can continue subject to fair-use limits. Extra local chats can use a separately billed API key.