Short answer: Start with ChatGPT Pro for the 1 to 5 people who will actually use Codex to do serious work. Move to ChatGPT Business when you need team governance, admin controls, offboarding, centralized billing, or stronger business privacy defaults. There is no separate Codex Pro subscription. Codex runs on your ChatGPT plan and is metered in credits.
If you are starting to use Codex as a business owner, the decision sounds like a product question. It is really a plan question: do you run Codex on a personal ChatGPT Pro plan, or on a ChatGPT Business seat?
One clarification first, because the labels trip people up. Codex does not have its own subscription. It is included with ChatGPT plans, and usage is metered in credits. So you are not choosing a Codex product. You are choosing which ChatGPT plan you run Codex on.
Best choice by business situation
| Situation | Best plan |
|---|---|
| Owner is the main AI user | ChatGPT Pro $100 |
| Owner or operator uses Codex daily | ChatGPT Pro $200 |
| 1 to 5 people do serious AI work | Individual Pro accounts |
| Whole team needs access | ChatGPT Business |
| Company needs admin controls or offboarding | ChatGPT Business |
| Company has compliance or privacy requirements | ChatGPT Business or Enterprise |
The simple rule
If only a few people will use Codex seriously, buy Pro for those people.
If the company needs managed team access, buy Business.
Do not buy 25 Business seats just because you have 25 employees. Buy capacity for the people who will turn AI into shipped work.
How Codex access works on ChatGPT Pro vs. Business
This is where the plan names confuse people, so here is the clean version.
On ChatGPT Pro (personal): Codex is included with your subscription and metered in credits. Pro has two tiers, $100 per month and $200 per month. Both include Codex. There is no standalone Codex Pro plan to buy.
On ChatGPT Business: there are two seat types, and the difference matters.
- Standard ChatGPT seats ($20 per user per month billed annually, $25 billed monthly, 2-seat minimum) include full ChatGPT workspace access plus baseline Codex access. Heavier Codex use draws on workspace credits.
- Codex seats (the Codex-only option) have no fixed monthly seat price. They are billed by usage through workspace credits, and they do not include ChatGPT workspace access.
A Business workspace can hold any mix of the two: only standard seats, only Codex seats, or both. OpenAI moved Codex seats to usage-based, token-metered pricing in April 2026.
So the real decision is not which Codex product. It is which ChatGPT plan you run Codex on, and whether your users need the full ChatGPT workspace or only Codex.
Why Pro usually wins for your first 1 to 5 users
On paper, ChatGPT Business looks cheaper. For a 25-person company, standard seats run about $500 per month billed annually or $625 billed monthly. Next to 25 Pro accounts, Business wins easily.
But that is the wrong comparison. Most small businesses do not have 25 people doing deep AI work. They have 1 to 5 people who actually turn AI into output: the owner, the operations lead, the marketing person, the assistant who handles follow-up, the website or technical person, the person responsible for client delivery. Those people need real capacity. Everyone else needs occasional access at most.
So the comparison that matters is a few Pro users against a full Business rollout:
| Setup | Monthly cost | What it means |
|---|---|---|
| 1 Pro user at $100 | $100 | One serious AI operator |
| 1 Pro user at $200 | $200 | One very heavy AI operator |
| 3 Pro users at $100 | $300 | Three core AI operators |
| 5 Pro users at $100 | $500 | Five serious AI operators |
| 3 Pro users at $200 | $600 | Three very heavy AI operators |
| 25 Business seats annual | $500 | Broad access for the whole team |
| 25 Business seats monthly | $625 | Broad access for the whole team |
If the company has 25 employees but only 3 will use Codex every week, three $100 Pro accounts cost $300 per month and put the money where the work actually happens.
This is where small businesses waste money. They buy seats for the whole team before they know who will use them. Two people adopt the tool, everyone else logs in once and goes back to old habits, and the company has AI access without AI leverage.
A better rollout:
- Pick the core AI users.
- Give them enough capacity.
- Assign real business outcomes.
- Measure what they produce.
- Expand only after the workflow is proven.
The people responsible for results get the horsepower first.
What Codex actually costs to run
Codex usage runs on credits. Once a plan's included credits are gone, heavier work draws on additional usage-based credits. That is where the cost gets serious.
OpenAI's current Codex rate card lists GPT-5.5 at:
| GPT-5.5 usage | Credits per 1M tokens | Dollar equivalent |
|---|---|---|
| Input tokens | 125 credits | $5 per 1M |
| Cached input tokens | 12.5 credits | $0.50 per 1M |
| Output tokens | 750 credits | $30 per 1M |
OpenAI's student Codex materials list 2,500 credits as $100, which works out to about $0.04 per credit. Treat that as an illustrative rate, since OpenAI does not publish one official dollar-per-credit number for general accounts. Using it, a $100 budget of GPT-5.5 usage roughly buys 20 million input tokens if it were all input, or 3.33 million output tokens if it were all output. At $200, that roughly doubles.
That sounds like a lot until you see how Codex works in real business tasks. It is not answering one short question. It reads project files, checks instructions, inspects a website, scans logs, edits files, tests changes, writes handoffs, uses tools, and holds context across a long assignment. That consumes far more than a simple chat.
OpenAI's own Codex rate card says average Codex usage runs about $100 to $200 per developer per month, with large variance depending on model, number of sessions, automations, and fast mode. So a serious Codex user lands in that $100 to $200 range either way. The difference is predictability. With personal Pro, you know the monthly price. With usage-based credits, the cost rises with how much serious work the person does.
When the $100 Pro plan makes sense
Use the $100 Pro plan for someone doing real work through the week, but not running the heaviest workflows all day: drafting content, preparing client materials, inspecting business files, organizing operating docs, running focused Codex tasks, supporting website and workflow improvements. For many owners, this is the right starting point. It is enough to stop treating Codex like a toy and start using it as a business tool.
When the $200 Pro plan makes sense
Use the $200 Pro plan when the person is becoming a true AI operator: using Codex most days, working across multiple projects, running longer sessions, inspecting real business systems, doing client delivery or revenue operations, and getting interrupted by usage limits. The $200 plan is not about prestige. It is about removing hesitation. If the person is asking "should I save my usage for later," the plan is already too small for the role.
When ChatGPT Business makes sense
This is not an argument against ChatGPT Business. Business is the right call when you need the business features: centralized billing, admin controls, managed team access, employee offboarding, workspace permissions, shared projects, shared company GPTs, app and connector controls, SAML or MFA, no-training-by-default for business data, and compliance support.
Those are real needs. Many small businesses just do not have all of them on day one. They need one person to turn AI into shipped work first.
Privacy and account sharing
One firm rule: do not share a single personal Pro login across the company. Each person should have their own account. The business can pay for or reimburse it, but the account belongs to one user.
Business also has the stronger default privacy posture. OpenAI says Business, Enterprise, Edu, and API inputs and outputs are not used to improve models by default. Consumer plans, including Plus and Pro, may have data used to improve models unless training is turned off in Data Controls. So if you have strict privacy, legal, compliance, or offboarding requirements, Business may be the right choice sooner. If you are owner-led, under 25 people, and trying to get practical work moving, Pro for your core users is the cleaner start.
FAQ
Is there a separate "Codex Pro" plan?
No. Codex does not have its own subscription. It is included with ChatGPT plans (Plus, Pro, Business, Enterprise) and metered in credits. ChatGPT Pro is the $100 or $200 personal plan, not a Codex product.
How much does ChatGPT Pro cost?
ChatGPT Pro has two tiers, $100 per month and $200 per month. Both include Codex.
How much does ChatGPT Business cost?
Standard ChatGPT Business seats are $20 per user per month billed annually, or $25 per user per month billed monthly, with a 2-seat minimum. Codex-only seats have no fixed monthly fee and are billed by usage through workspace credits.
Does ChatGPT Business include Codex?
Yes. Standard Business seats include baseline Codex access along with full ChatGPT workspace access, and heavier Codex use draws on workspace credits. Codex-only seats are separate, billed by usage, and do not include ChatGPT workspace access.
Should a 25-person company buy 25 Business seats?
Usually not. Most small companies have only 1 to 5 people who will use Codex seriously. Put those people on Pro first, then move to Business when you need team governance.
What does it cost to run Codex each month?
OpenAI's Codex rate card cites roughly $100 to $200 per developer per month, depending on model, number of sessions, automations, and fast mode.
Is my data used to train OpenAI's models?
On ChatGPT Business, Enterprise, Edu, and the API, inputs and outputs are not used to train models by default. On consumer plans such as Plus and Pro, data may be used to improve models unless you turn training off in Data Controls.
Which is more private, ChatGPT Pro or ChatGPT Business?
Business has the stronger default privacy posture, because business data is not used for model training by default.
The bottom line
For small business teams of 25 or fewer, the best first AI investment is usually not broad access. It is concentrated execution. Put the owner, operator, marketer, assistant, or delivery lead on the plan that lets them actually use Codex and ChatGPT to produce business outcomes. Once the company needs shared controls and team governance, move to Business. Do not start by buying seats for people who are not ready to use them. Start with the people who will turn AI into work.
Sources
- OpenAI: ChatGPT Business pricing
- OpenAI Help Center: What is ChatGPT Business (seat types, Codex access)
- OpenAI Help Center: Managing billing and seats in ChatGPT Business
- OpenAI: Codex pay-as-you-go pricing for teams
- OpenAI Help Center: About ChatGPT Pro tiers
- OpenAI Help Center: Codex rate card
- OpenAI Developers: Codex pricing