ChatGPT Enterprise pricing 2026: seat, add-on & API cost.
A buyer-side reference for procurement, IT and finance leaders negotiating ChatGPT Enterprise. This note sets out per-seat tiers, Sora and Operator add-ons, API token cost, EU, Canadian and Japanese data-residency premiums, realised discount bands by seat count, and the all-in cost model for 1,000 to 10,000 seat rollouts — the Enterprise contract, not the consumer Team or Plus plans.
ChatGPT Enterprise lists at $60 per seat per month on annual commit and $80 on monthly, but realises at $40–$52 above 2,500 seats on three-year terms. The cost buyers underestimate is the add-on and API layer — Sora, Operator, data residency and token spend. For a 1,000-seat rollout, plan for $620K–$980K per year all-in.
01 Key findings
List price is a ceiling, not the deal. Enterprise lists at $60 per seat per month (annual) or $80 (monthly); realised pricing runs $40–$52 per seat above 2,500 seats on three-year terms. Team is $25, Business $30, Edu $10 for accredited institutions.
The add-on layer is the hidden cost. Sora video generation is $20 per user per month for 500 seconds; Operator agentic browsing is $25 per user per month; data residency adds 8–15%. License these to named groups, not every seat.
Only Enterprise clears the security bar. Business is the cheapest tier meeting baseline controls, but Enterprise alone delivers default zero-retention, full audit, HIPAA BAA and indemnity language legal teams require for high-risk work.
API token spend runs separately and rivals seat cost. Embedded-app consumption is billed per token on top of seats; batch and cached input each cut 50%, and Azure PTUs land 30–45% below per-token at sustained load.
Discount realisation tracks leverage, not list. Seat count, three-year term and a credible Claude Enterprise or Microsoft 365 Copilot alternative are what move OpenAI — typically 15–28% off the opening quote.
02 Enterprise 2026 seat pricing
OpenAI publishes ChatGPT seat pricing across five tiers in 2026: Free, Plus ($20 per user per month), Team ($25–$30), Business ($30), Enterprise (custom, with $60–$80 per seat as the typical reference), and Edu ($10 for accredited institutions). Enterprise is the only tier that includes SOC 2 Type II controls, SAML SSO, custom retention windows, and the contractual zero-retention commitment most enterprise buyers require.
| Tier | List per user per month | Annual commit per user |
|---|---|---|
| ChatGPT Free | $0 | $0 |
| ChatGPT Plus | $20 | $240 |
| ChatGPT Team (annual) | $25 | $300 |
| ChatGPT Team (monthly) | $30 | $360 |
| ChatGPT Business (annual) | $30 | $360 |
| ChatGPT Enterprise (annual) | $60 to $80 | $720 to $960 |
| ChatGPT Edu | $10 (accredited institutions) | $120 |
A 1,000-seat Enterprise rollout at the $60 reference lists at $720,000 per year. Realised pricing on a three-year commit at 1,000 seats typically lands between $42 and $52 per seat per month, or $504,000 to $624,000 per year. The realised tier depends on TCV, multi-year commit length, and the credible threat of Claude Enterprise or Microsoft 365 Copilot displacement.
The Plus-to-Enterprise upgrade anchor: enterprises that start with employees expensing ChatGPT Plus individually often find OpenAI sales anchoring the consolidation at the Plus price times 3x. The counter: Plus contains none of the enterprise controls (no SSO, no zero-retention, no admin console, no audit), so the Enterprise unit price must be evaluated on its own feature merits, not as a Plus multiplier. See our OpenAI Enterprise pricing analysis.
03 Enterprise vs Business vs Team vs Edu
The tier decision determines what the customer gets for the money. Enterprise is the only tier with the controls that most security, legal and privacy teams require for production use.
| Feature | Team | Business | Enterprise |
|---|---|---|---|
| SAML Single Sign-On | No | Yes | Yes |
| SCIM provisioning | No | Yes | Yes |
| Custom data retention | No | Limited | Yes (configurable) |
| Zero-retention commitment | No | No | Yes (default) |
| Admin console + audit log | Basic | Yes | Yes (full) |
| Compliance API (HIPAA BAA) | No | No | Yes |
| Custom GPT distribution | Team-only | Org-wide | Org-wide + sharing controls |
| GPT-4 / GPT-4o / o-series access | Yes (rate-limited) | Yes | Yes (unlimited) |
| Sora video generation | Limited | Add-on | Add-on |
| Operator agentic browsing | No | Add-on | Add-on |
| Priority support | No | Standard | Premium SLA |
Business is the cheapest tier that meets baseline security requirements, but Enterprise is the only tier that delivers the audit, retention and indemnity language legal teams require for production LLM deployment. The right structure for most organisations is to deploy Business to broad-population low-risk users (sales, marketing, general office) and Enterprise to high-risk knowledge work (legal, finance, engineering, research).
04 Sora & Operator add-ons
Sora is OpenAI's text-to-video model. It is bundled into ChatGPT Pro for individuals at $200 per month and sold as an add-on to ChatGPT Enterprise at $20 per user per month for 500 seconds of generation. Heavy generation over the included pool is billed at $0.50 per second of video. Sora is consumption-priced under a per-seat envelope, so the right structure is to license it to a small named group (creative, marketing video, internal communications, sales enablement) rather than every seat. Blended Sora attach across our 2026 rollouts sits at 8–14% of Enterprise seats.
Operator is OpenAI's agentic browsing platform: the agent operates web applications, fills forms and executes multi-step web tasks under user supervision. Operator is $25 per user per month, bundled with ChatGPT Pro and sold as an Enterprise add-on. Use cases are narrow but high-value — travel booking, expense submission, vendor-portal data entry, competitive research. Attach rates sit at 5–12% of Enterprise seats; generic "agent for everyone" deployments do not deliver measurable productivity gains and produce visible budget overruns.
05 OpenAI API token pricing
Most ChatGPT Enterprise customers also consume the OpenAI API for embedded AI features in custom applications. API pricing is per-token and runs separately from ChatGPT seat pricing. The 2026 token prices for the production model families are below.
| Model | Input per 1M tokens | Output per 1M tokens |
|---|---|---|
| GPT-4o | $2.50 | $10.00 |
| GPT-4o mini | $0.15 | $0.60 |
| GPT-4.1 | $2.00 | $8.00 |
| GPT-4.1 mini | $0.40 | $1.60 |
| o1 (reasoning) | $15.00 | $60.00 |
| o1-mini | $3.00 | $12.00 |
| o3-mini (reasoning) | $1.10 | $4.40 |
| text-embedding-3-large | $0.13 | n/a (embeddings) |
| text-embedding-3-small | $0.02 | n/a (embeddings) |
Batch API pricing runs at 50% of standard for non-real-time workloads, and cached input runs at 50% of standard input. Provisioned Throughput Units (PTU) on Azure OpenAI provide reserved capacity at a per-PTU monthly charge that frequently lands 30–45% below per-token billing at sustained throughput above 100 tokens per second. See our enterprise LLM cost comparison for the cross-vendor token economics.
06 Data residency & enterprise controls
ChatGPT Enterprise added EU data residency in 2024, and Canadian and Japanese residency through 2025 and 2026. Residency commits OpenAI to processing and storing customer data within the selected geography, addressing GDPR, Canadian PIPEDA and Japanese APPI requirements. It is configured per Enterprise tenant and applies to chat data, file uploads and custom GPT data. The premium is roughly 8–15% above the standard Enterprise unit price — non-negotiable for organisations subject to EU data-localisation rules, and otherwise weighed against the legal team's actual data classification.
The Enterprise controls that justify the unit price are: SAML 2.0 SSO with major identity providers (Okta, Entra ID, Ping), SCIM 2.0 automated provisioning and deprovisioning, configurable retention from one day to indefinite, default zero-retention for input and output (no training on customer data), full admin console with usage analytics, audit log with API export, HIPAA BAA availability, SOC 2 Type II, ISO 27001, and the OpenAI Compliance API for security-tool integration.
The sub-processor question: OpenAI's DPA names Microsoft Azure as a compute sub-processor. If Microsoft is on the customer's no-sub-processor list, OpenAI Enterprise on the standard infrastructure is not viable — the alternative is Azure OpenAI Service through the customer's own Microsoft contract, which collapses the concern into the existing relationship. Pair this with the seat-floor trap: minimum-commit thresholds on Enterprise and residency tenants can strand under-utilised seats, so commit at 60–70% of forecast and add on windows. See our AI contract data residency and IP rights guide.
07 Realised discount bands by seat count
Discount realisation varies primarily by seat count, term length, and the credible competitive alternative. The benchmarks below reflect typical realised discounts observed in advisor-led ChatGPT Enterprise negotiations during 2024 to 2026.
| Seat count band | Annual commit price | Three-year commit price |
|---|---|---|
| 250 to 500 seats | $55 to $65 per seat per month | $48 to $58 |
| 500 to 1,500 seats | $48 to $58 | $42 to $52 |
| 1,500 to 5,000 seats | $42 to $52 | $36 to $46 |
| 5,000 to 15,000 seats | $36 to $46 | $30 to $42 |
| 15,000+ seats | $30 to $40 | $25 to $36 |
08 1,000 & 10,000-seat cost model
The all-in annual cost combines seat licence cost, add-on attach (Sora, Operator), the data-residency premium where applicable, and the API consumption layer for embedded applications. The two reference rollouts below illustrate realistic 2026 commercial outcomes.
| Cost line | 1,000-seat rollout | 10,000-seat rollout |
|---|---|---|
| Enterprise seat licences (3yr commit) | $540,000 | $3,600,000 |
| Sora add-on (10% attach) | $24,000 | $240,000 |
| Operator add-on (8% attach) | $24,000 | $240,000 |
| EU data residency premium (assume 25% in EU) | $20,000 | $200,000 |
| API token consumption (embedded apps) | $60,000 to $180,000 | $400,000 to $1,500,000 |
| Total annual | $668K to $788K | $4.68M to $5.78M |
Realised three-year per-seat pricing falls sharply as seat count rises — the single largest lever on the seat line of the model above:
09 Negotiation levers that work in 2026
Five levers move OpenAI account teams. Weight them to your seat count and competitive position before opening the conversation.
Credible Claude quote in hand
OpenAI flexibility correlates directly with documented Anthropic alternative pricing. See our Claude Enterprise pricing pillar.
Copilot as displacement
For Microsoft-first estates, Microsoft 365 Copilot at $30 per seat (on top of E3 or E5) is the natural cost comparator. See Copilot pricing 2026.
Multi-year for unit relief
OpenAI accepts material unit-price relief in exchange for three-year terms versus annual — the cleanest single trade in the deal.
Phased, milestone seat adds
Commit at 60–70% of forecast initial seats with quarterly add windows at the contracted unit price, protecting against seat-floor overspend.
Bundle API with seats
Account teams have authority to discount API consumption in exchange for seat commit, especially for embedded application use cases. See the OpenAI negotiation white paper.
10 Alternatives & recommendation
The ChatGPT Enterprise decision is rarely standalone. Most 2026 buyers evaluate at least two of ChatGPT Enterprise, Claude Enterprise, Gemini Enterprise and Microsoft 365 Copilot. The criteria that matter are model capability for the customer's actual workloads, contract terms (especially indemnity and data residency), and integration with existing collaboration and identity infrastructure.
The strongest displacement quote to hold. Weigh indemnity terms and model fit for your workloads. See Claude Enterprise pricing 2026.
Natural for Workspace-centric estates that can fold AI into the existing Google contract. See Gemini Enterprise pricing.
$30 per seat on top of E3 or E5, with the tightest Office integration. See Copilot pricing 2026 and the Copilot rollout guide.
The full vendor decision framework lives in our enterprise AI vendor selection framework. For broader counsel see AI procurement advisory, the AI procurement guide, software licensing advisory, and cloud contract negotiation.
Negotiate ChatGPT Enterprise with the full picture
Independent reviews land a median 22% below OpenAI's opening quote. We run the negotiation in shadow advisory or on your behalf, on fixed fee.
The Licensing Edge
Weekly vendor and AI-licensing intelligence for enterprise IT leaders. 3,000+ subscribers.