THINXSTER
Blog/AI Marketing
AI Marketing7 min readAugust 16, 2026

OpenAI Enterprise Cost: ChatGPT Pricing Explained

ChatGPT Enterprise isn't publicly priced. Buyers report ~$60/user/month, a 150-seat minimum and 12-month term — roughly $108,000/year before discounts.

RK
Ryan Korsz
Founder & CEO, Thinxster

TL;DR

ChatGPT Enterprise isn't publicly priced. Buyers report ~$60/user/month, a 150-seat minimum and 12-month term — roughly $108,000/year before discounts.

→ See how this applies to your business (free 30-min call)

OpenAI does not publish a price for ChatGPT Enterprise. The figure consistently reported by buyers and press since launch is roughly $60 per user per month, with a 150-seat minimum and a 12-month commitment — a floor of about $108,000 per year before you have used it once. Volume and multi-year terms pull the effective seat price to the $40–$50 range; a few large logos have reported lower. Everything below that tier is publicly priced: ChatGPT Business runs $30/user/month billed monthly or $25 billed annually, ChatGPT Plus is $20/month, and API access is billed separately per token. Verify current numbers with OpenAI sales — this contract is negotiated, not listed.

What ChatGPT Enterprise Actually Costs

The seat price is the smallest part of the conversation. Three contract terms drive the real number:

  • Seat minimum. 150 seats is the commonly cited floor. If your company has 60 employees, you are buying 90 seats of nothing. At $60, that is $64,800 per year of pure shelfware.
  • Term length. 12 months is standard; sales will push 24 or 36 for a discount of roughly 10–20%. Given how fast model pricing has fallen, a 36-month lock is usually a bad trade (more on that below).
  • Credits vs. seats. Enterprise agreements increasingly bundle API credits or "compute credits" alongside seats. Credits that expire annually are a use-it-or-lose-it liability, not a discount.
  • A realistic 200-seat Enterprise deal lands between $120,000 and $144,000 per year. Add a $15,000–$40,000 internal implementation effort in year one — training, policy, IT integration, someone owning it — and the true first-year number for a mid-market company is closer to $150,000–$180,000.

    The Cheaper Tiers Most Service Businesses Should Price First

    If you run an HVAC company with 40 office staff, a multi-location dental group, or a 90-person law firm, ChatGPT Enterprise is almost certainly the wrong SKU. The math:

  • 40 seats on ChatGPT Business, annual billing: 40 × $25 × 12 = $12,000/year.
  • The same 40 people on Enterprise (150-seat minimum): 150 × $60 × 12 = $108,000/year.
  • Difference: $96,000/year, or 9x, for a company that would use fewer than a third of the seats it paid for.
  • What you actually give up at the Business tier is a specific, checkable list: no SAML SSO in some configurations (SCIM provisioning and domain verification vary by tier), no enterprise-grade admin analytics, no expanded context windows on certain models, no dedicated account team, and no negotiated data-residency or custom retention terms. Business already includes the thing most buyers think they are paying Enterprise for: your data is not used to train models by default on all business tiers. That one point is worth confirming in writing, because it is the reason a large share of Enterprise deals get approved in the first place — and it is frequently not the differentiator the room assumes.

    The most common OpenAI Enterprise mistake is buying 150 seats to solve a compliance question that a $25/seat plan and a two-page acceptable-use policy would have answered.

    API Pricing Is a Completely Different Cost Model

    Seats price *people*. The API prices *work*. Recent list pricing for flagship models has run roughly $1.25–$2.50 per million input tokens and $8–$10 per million output tokens, with smaller models at $0.15–$0.60 per million input. Two discounts matter and are routinely missed:

  • Prompt caching cuts repeated input tokens by 50–90%. If you send the same 4,000-token system prompt on every call, this is the single highest-ROI change you can make.
  • The Batch API takes 50% off for jobs you can wait up to 24 hours on — nightly review summarization, bulk content generation, backlog classification.
  • Concrete workload math: a service business generating 2,000 customer-service reply drafts per month, at ~1,500 input and ~400 output tokens each, burns about 3M input and 0.8M output tokens. At $2.50/$10, that is roughly $15.50 per month. With caching, under $9. That same workload, sold as 20 Enterprise seats, would cost $1,200 per month.

    This is the number that reframes the whole purchase. For narrow, repeatable, high-volume tasks, the API is 50–100x cheaper than seats. For open-ended knowledge work by humans who need a chat interface, seats win, because building and maintaining an interface costs more than $25/person/month. Most companies need both, in different proportions than they expect.

    The Line Items Nobody Puts in the Quote

  • Identity tier upgrades. SSO/SCIM often forces an upgrade on your identity provider. Okta or Entra ID premium tiers commonly add $3–$8 per user per month — on 200 users, another $7,200–$19,200/year.
  • Security review. A vendor security assessment, DPA negotiation, and legal review runs 20–60 hours of internal time. At a blended $120/hour, that is $2,400–$7,200.
  • The enablement gap. Untrained seats go dark. Internal usage telemetry across rollouts consistently shows 30–50% of licensed seats become inactive by month 3 without a structured enablement program. On a $108,000 contract, 40% dormancy is $43,200 wasted.
  • Shadow spend. Teams that already pay for Copilot, Gemini, Claude, Jasper, and three point solutions frequently end up with 2–4 overlapping AI subscriptions per department after an Enterprise rollout, not fewer.
  • Price deflation risk. Cost per unit of equivalent model capability has fallen by well over 10x in roughly 24 months. A 36-month lock at today's rate assumes that stops. It has not stopped yet.
  • When ChatGPT Enterprise Is a Bad Purchase

    This is the part the sales deck skips. Do not buy it if any of these describe you:

  • You have fewer than 150 employees. You will pay for seats you cannot fill. Business tier at $25–$30 is the honest answer, and no amount of negotiation fixes a minimum that is 3x your headcount.
  • Your use case is one workflow. If the real goal is "summarize call transcripts" or "draft estimate follow-ups," you want an API integration costing $50–$400/month, not a $9,000/month seat contract. Buying seats to solve a workflow problem is the most expensive mistake in this category.
  • You have no owner. If nobody's job description includes AI adoption, expect sub-25% weekly active usage by month 6. The tool does not adopt itself, and a $108,000 line item with 25% engagement is a $81,000 write-off that will poison the next three AI proposals internally.
  • Your data is a mess. Enterprise features like company knowledge connectors only help if your documents are organized, current, and permissioned. Pointing a model at a SharePoint with seven versions of the 2023 price list produces confident wrong answers faster than a human could.
  • Your regulated requirements exceed what a standard contract covers. HIPAA BAAs, FedRAMP, specific data residency, and long retention holds are all negotiable but not automatic — and if the answer comes back "not supported for your configuration," you have spent three months to learn it.
  • You are buying it to look modern. Board-driven AI purchases have the worst usage curves of any category we see.
  • The failure mode we watch most often is subtler than any of these: a company buys Enterprise, gets genuine value from 8–12 power users, and never notices that those 12 people are producing all the ROI while 188 seats idle. That is a real win hiding inside a bad contract. The fix is a smaller seat count plus API-backed automation for everyone else, and it usually cuts spend 60–75% with no loss of output.

    How to Model It Before You Sign

    Run this in a spreadsheet before the first sales call, not after:

    1.

    Count people who would use it weekly, not people with a login. Multiply by $25 (Business) and by $60 (Enterprise, or 150 × $60 if you are under the minimum).

    2.

    List your top five repeatable tasks. Estimate volume per month and tokens per task. Price them at API rates. Anything with more than 500 monthly repetitions belongs in automation, not in a chat window.

    3.

    Assign a dollar value to hours saved. If a $28/hour CSR saves 40 minutes a day, that is roughly $243/month — one seat pays for itself nearly 10x at Business pricing and 4x at Enterprise pricing, *if* the time saved is real and reallocated.

    4.

    Put a 40% dormancy haircut on your seat count. If the deal still works, it is a good deal. Our ROI calculator handles this math if you would rather not build it.

    Step 4 is the one that kills most deals, and it should.

    What We Tell Service Businesses

    For the plumbing, HVAC, dental, legal, and home services companies we work with — typically 10 to 200 employees — the pattern that works is boring and cheap: 8–20 Business seats at $25/month for the people who genuinely think in text all day, plus API-backed automations for intake, follow-up, review requests, and estimate drafting that run at $100–$600/month in token costs regardless of headcount. Total: often under $1,200/month, against a $9,000/month Enterprise floor.

    Enterprise makes sense above roughly 250 real weekly users, or when a specific compliance requirement is genuinely unmet by lower tiers and you have the letter from legal to prove it. Below that, you are buying a procurement badge.

    If you want the workflow-level version of this analysis for your business, our pricing page shows what implementation actually costs, and the free marketing audit includes a usage-and-spend model built on your headcount and ticket volume rather than a vendor's minimum.

    Frequently Asked Questions

    How much does ChatGPT Enterprise cost per user?

    OpenAI does not publish a list price. Buyers and press consistently report roughly $60 per user per month at the standard tier. Volume commitments and multi-year terms typically bring the effective seat price down to the $40–$50 range, and some large accounts have reported paying less.

    What is the minimum number of seats for ChatGPT Enterprise?

    The commonly cited floor is 150 seats, paired with a 12-month commitment. At roughly $60 per seat per month, that puts the practical entry point near $108,000 per year. Because the contract is negotiated rather than listed, both the seat minimum and term are discussion points with OpenAI sales.

    What's the difference between ChatGPT Business and Enterprise pricing?

    ChatGPT Business is publicly priced at $30 per user per month billed monthly, or $25 per user per month billed annually, with no large seat minimum. Enterprise is unlisted, negotiated, and starts around $60 per seat per month with a 150-seat minimum and an annual commitment.

    Does ChatGPT Enterprise include API access?

    No. API usage is billed separately on a per-token basis and is not covered by Enterprise seat licenses. Organizations that build their own applications on OpenAI models pay those token costs on top of seat costs, so budget the two line items independently.

    Free Weekly Briefing

    One AI Marketing Tactic.
    Every Tuesday. Free.

    What's actually working across our client accounts right now — ROAS moves, follow-up sequences, creative angles. The stuff that isn't in any blog post yet.

    No spam. Unsubscribe anytime. 1,200+ business owners already in.

    Ready to Deploy

    SEE THIS IN
    YOUR BUSINESS.

    30 minutes. We scope the exact systems that apply to your situation and give you a plan.

    ★★★★★ Trusted by 47+ local service businesses

    BOOK A STRATEGY CALL →