ChatGPT Enterprise Pricing: What It Costs and How Seats Work

How much does ChatGPT Enterprise cost?

ChatGPT Enterprise pricing is quote-only: the plan has no list price. OpenAI's pricing page shows the Enterprise card with "Custom pricing" and a contact-sales button, and the FAQ directs buyers to sales rather than quoting a figure. Anyone stating a definitive per-seat number for ChatGPT Enterprise is reporting what some buyers negotiated, not an official rate.

What OpenAI does publish is the tier immediately below it. ChatGPT Business is $20 per user per month billed annually, or $25 per user per month billed monthly, available from two users. That figure is the useful anchor: it is the floor your Enterprise quote will sit above, and for a large share of companies asking about Enterprise pricing, Business is the plan they actually want.

The absence of a public price is deliberate rather than evasive. Enterprise agreements bundle custom data retention, data residency, SLAs, custom legal terms, and volume discounts, all of which vary by customer. A single list price cannot express that, so OpenAI prices it per deal. Practically, that means your negotiating room as a buyer is real, and the number you are quoted first is rarely the number you end up paying.

What OpenAI publishes about ChatGPT Enterprise pricing

Separating published fact from market rumour is the whole job on this topic, so here is the published side. OpenAI states that paid plans are priced per user per month, that monthly billing is available for Go, Plus, and Business, and that annual plans are offered for Business and Enterprise. Business starts at two users. Enterprise is contact-sales only.

Two published details cut real money off the price and get overlooked. OpenAI states that nonprofits can access up to a 75% discount on ChatGPT Business or ChatGPT Enterprise through its nonprofit program. And it offers ChatGPT Edu for universities plus a free ChatGPT for Teachers plan for verified US K-12 educators through June 2027. If you are in either category, the commercial price list is not your price list.

OpenAI also notes that Enterprise and Business customers can purchase credits for additional access, which matters for capacity planning: heavy usage is not purely a function of seat count. And Enterprise buyers get alternative payment options such as invoicing, where the self-serve tiers require a credit card.

• All paid plans are priced per user per month; annual billing exists for Business and Enterprise.
• Business starts at two users, so there is no large minimum before a team can buy.
• Nonprofits can access up to a 75% discount on Business or Enterprise via OpenAI's nonprofit program.
• Business and Enterprise can buy credits for additional access beyond what seats include.
• Enterprise supports invoicing; self-serve tiers require a card.

How seat-based pricing works

Seat-based pricing means you pay per named user per month, regardless of how much any individual uses the product. Fifty seats costs fifty times one seat. The model is simple to forecast and simple to over-buy, and the second of those is where most of the wasted spend in enterprise AI sits.

The failure pattern is consistent and worth planning against. A company buys seats for a whole department because it feels equitable, adoption lands somewhere between a third and a half of those people, and the rest of the licences sit idle for a full annual term. Because seats are usually committed annually, you cannot true down mid-term. The money is spent whether it is used or not.

The defence is unglamorous: buy for demonstrated usage, not headcount. Start with a pilot group who have volunteered, watch weekly active use for a quarter, and expand from evidence. If your vendor's admin console gives you usage analytics, which both Business and Enterprise do, that is exactly the data this decision needs. A 40-seat contract that is 90% used beats a 100-seat contract that is 35% used, at almost any per-seat rate.

What ChatGPT Enterprise includes

Enterprise is not a better model, it is better governance. OpenAI's Enterprise card lists an expanded context window supporting longer inputs and larger files, enterprise-level security and controls including SCIM, enterprise key management, user analytics, domain verification and role-based access controls, advanced data privacy with custom data retention policies, encryption at rest and in transit, and no training on business data by default.

Beyond security, the differentiators are operational: support for data residency in ten regions, 24/7 priority support with SLAs, custom legal terms, access to OpenAI's AI advisors for eligible customers, and invoicing with volume discounts. That list reads like a procurement checklist because it is one. These are the items that block deployment in regulated industries.

Note what is not on the list: a smarter model. The published feature comparison shows the same model families across paid tiers, with Enterprise adding capacity and controls rather than intelligence. If the reason someone wants Enterprise is better answers, the upgrade will disappoint. If the reason is that legal will not approve anything without a DPA, custom retention, and audit logging, it is exactly the right purchase.

• Expanded context window for longer inputs and larger files than the self-serve tiers.
• SCIM provisioning, enterprise key management, domain verification, and role-based access controls.
• Custom data retention policies, encryption at rest and in transit, no training on your business data by default.
• Data residency support across ten regions.
• 24/7 priority support, SLAs, custom legal terms, and access to AI advisors for eligible customers.
• Invoicing and volume discounts rather than card-only self-serve billing.

Enterprise vs Business vs Plus

Most companies asking about ChatGPT Enterprise pricing should buy Business. Business already includes a secure workspace with SAML SSO and MFA, centralized billing and administration, usage analytics with budgeting and spend controls, connectors to tools like Microsoft 365, Google Drive, Slack, GitHub, Linear and Figma, custom team agent plugins built on company knowledge, and no training on business data by default, all at a published $20 per user per month annually.

The jump to Enterprise buys four things Business does not have: SCIM and enterprise key management, custom data retention and data residency, contractual guarantees in the form of SLAs, custom legal terms, and named support, and an expanded context window. If none of those four appear in your requirements document, you are looking at a negotiation you do not need to have.

Plus at $20 a month is a personal subscription and belongs in a different conversation. Ten people expensing Plus individually costs roughly the same as ten Business seats and gives you no admin console, no unified billing, no shared workspace, and no contractual position on training data. Shadow AI spend on personal Plus accounts is a common and quietly expensive pattern; consolidating it onto Business usually costs nothing extra and fixes the governance problem outright.

What buyers report paying

Because OpenAI publishes nothing, the only figures circulating are buyer reports and analyst estimates, and they should be read as exactly that. Across procurement write-ups and negotiation advisories in 2026, reported ChatGPT Enterprise contracts cluster in a range roughly between $45 and $75 per user per month, with larger deployments and longer commitments landing at the lower end. Several sources also describe a seat minimum in the low hundreds and annual prepayment as standard.

Treat all of that as directional, not authoritative. These numbers are not confirmed by OpenAI, they vary by source, they change as OpenAI restructures its plans, and any given quote depends on seat count, term length, region, and what else you are buying. Building a budget on a third-party per-seat figure and then discovering your quote is 40% higher is a bad way to start a procurement cycle.

The defensible way to use these reports is as a negotiation reference point rather than a forecast. Knowing that other buyers have landed in a range tells you whether your quote is unusual. It does not tell you what you will pay. Get the quote, then compare it against your Business-tier alternative at a published $20 per user per month, because that comparison is the one that has a real number on both sides.

Costs beyond the seat price

The seat rate is rarely the whole bill. Four other line items show up on enterprise AI deployments, and they are the ones that turn a comfortable business case into an uncomfortable one at renewal.

The largest is almost always unused seats. At a reported $50 to $60 per seat per month, a 200-seat contract running at 40% adoption is wasting somewhere near $70,000 a year on licences nobody opens. That single number usually dwarfs every other overhead in the deployment, which is why usage analytics are worth more than they look on a feature list.

The others are enablement, integration, and the API. Rolling out an assistant to a few hundred people means training, internal documentation, and someone owning adoption, none of which is on the invoice. Connecting it to internal systems is engineering time. And building anything customer-facing means API spend, which is billed entirely separately from ChatGPT seats and does not draw down against them.

• Unused seats, typically the single largest waste line in per-seat AI contracts.
• Enablement and training, which determine adoption and therefore the real cost per active user.
• Integration engineering to connect the assistant to internal systems and data sources.
• API spend for anything customer-facing, billed separately from seats with no cross-credit.
• Additional credits, which OpenAI notes Business and Enterprise customers can purchase for more access.
• Annual prepayment, which removes the option to true down mid-term as adoption becomes clear.

Do Enterprise seats include API access?

No. ChatGPT seats and the OpenAI API are separate products on separate meters, and this is the single most common budgeting mistake companies make when they start building rather than just using. A ChatGPT Enterprise contract licences named humans to use the ChatGPT interface. It does not grant your engineering team API credits, and API spend does not draw down against your seats.

The distinction is worth internalizing because the two are priced on completely different logic. Seats are priced per person per month, flat, regardless of how much any individual types. API access is priced per token, which means it scales with what your software actually processes rather than with headcount. Neither is better; they answer different questions.

Where this bites is the moment a company decides to build something customer-facing on top of the assistant it already pays for. The finance conversation goes badly, because the existing contract covers none of it and the new line item is usage-based rather than predictable. Plan the two budgets separately from the start, and be clear internally about which projects are seat-funded and which are token-funded.

It also explains why the seat price is a poor guide to what an AI product costs to run. Raw inference on a modern frontier model is cheap per answer, often fractions of a cent. The expensive parts of shipping something customer-facing are the integrations, the grounding, the escalation rules, and the operational ownership. None of those appear on a per-seat price list.

When ChatGPT Enterprise is worth it

Enterprise is worth it when compliance requirements make the cheaper tiers unusable, and rarely otherwise. The test is not company size or budget; it is whether specific items on the Enterprise list are hard requirements you cannot deploy without.

1. 1You need SCIM provisioning because manually managing hundreds of accounts across joiners and leavers is not viable and creates a security gap.
2. 2You have a regulatory data residency obligation that requires processing in a specific region, which Enterprise supports across ten regions and Business does not.
3. 3Legal will not sign without custom data retention terms, custom legal terms, or a contractual SLA.
4. 4You need enterprise key management to satisfy an internal security standard on encryption control.
5. 5You are large enough that volume discounting on a negotiated contract genuinely beats the published $20 per user per month, which starts to bite at scale.
6. 6Your workloads need the expanded context window for consistently large documents, and you have tested that the Business tier's limits actually constrain you.

How to negotiate an Enterprise contract

Quote-only pricing cuts both ways. It means you cannot budget from a website, and it means the price is negotiable in a way a published rate is not. The buyers who do best treat it like any other enterprise software negotiation rather than like a subscription purchase.

1. 1Establish your alternative first. ChatGPT Business at a published $20 per user per month is a real, credible fallback, and knowing exactly what you lose by choosing it is the strongest position you can hold.
2. 2Right-size before you quote. Run a pilot, measure weekly active usage, and ask for the seat count you can evidence rather than the one on the org chart.
3. 3Ask for a usage-based true-up rather than an upfront commitment on your full projected headcount, so growth is priced later rather than paid for now.
4. 4Put the non-price terms on the table too: retention periods, residency region, SLA specifics, security review scope, and onboarding support all carry real value.
5. 5Check whether you qualify for a program rate. OpenAI publishes up to a 75% nonprofit discount and separate education plans, and those are not discretionary discounts you have to argue for.
6. 6Time it. Vendor quarter-end is a real factor in enterprise software, and a deal that is ready to sign in the last two weeks of a quarter tends to be a cheaper deal.
7. 7Agree the review checkpoint in the contract. A written six-month usage review with an agreed mechanism for adjustment is worth more than a few dollars off the seat rate.

Per-seat AI vs support automation pricing

Per-seat pricing is the correct model for what ChatGPT Enterprise is. Value from an internal assistant scales with the number of employees using it, so a price that scales with employees tracks the value. Two hundred people writing, analyzing, and coding faster is roughly two hundred times one person doing so, and a per-user rate expresses that cleanly.

Customer support does not work that way, and this is where companies get the model wrong. Support value scales with the number of customer conversations handled, not the number of staff logged in. A store with three support people and fifty thousand orders a month does not have a headcount problem that three more AI seats would fix. It has a volume problem, and per-seat pricing cannot price volume.

There is a capability gap on top of the pricing mismatch. An internal assistant cannot look up order 10482, cannot apply a return window to a specific purchase date, cannot issue a refund inside merchant-set caps, and has no defined moment where it stops and hands a customer to a human with full context. Those are integrations and rules rather than model capabilities, which is why buying a higher tier of a general-purpose assistant never closes the gap.

Bookbag prices support automation the way support actually behaves: flat monthly plans with message-credit allowances and a spend cap the merchant sets, not per-seat licences. One credit equals one AI reply on any model, and a typical conversation runs about four replies, so a plan maps to conversation volume rather than headcount. Free covers 50 credits, Starter is $30 a month for 600, and Growth is $110 a month for 5,000. The agent connects to Shopify, WooCommerce, or BigCommerce so order questions become live lookups, and hands off to a human with context when a question falls outside what the data supports.

One model to avoid on this side of the market: per-resolution billing, where the invoice rises the more effectively the automation performs. It is the pricing structure merchants complain about most with general chatbot platforms, because it points the vendor's incentives away from the buyer's. Flat plans with credits and a spend cap keep peak season predictable, which matters more over a year than any headline monthly rate.