Enterprise AI Budget Planner
Illustrative annual scenarios for Claude and ChatGPT: seat fees, assumed usage and estimated seat commitments. Contract terms and actual usage may differ.
| Tier | Seat price | Usage and allowance | Contract terms | |
|---|---|---|---|---|
| Claude Team 2 – 150 |
$20 Standard · $100 Premium (annual); $25 · $125 monthly | Included per seat, on a rolling 5-hour window plus weekly limits. Premium is 5x Standard. Usage credits past the limit at API rates; pooled API credits up to $500/month. | Monthly or annual, card only. Seats can be added mid-term. | Source |
| Claude Enterprise 20 minimum self-serve; 50 sales-assisted |
$20 per seat per month, billed annually | None included. All usage billed at standard API rates on top of the seat, no seat-level limits; admins set spend caps per org or user. | Annual only. Seats billed at the start of the year, additions prorated; self-serve seats cannot be removed mid-term. Self-serve usage is prepaid credit and stops at zero; sales-assisted usage is invoiced monthly in arrears. | Source |
| ChatGPT Business 2 or more |
$20 per user per month annual; $25 monthly | Codex included at roughly Plus-level limits (GPT-6.1 Sol: 15 – 160 local messages per 5 hours). Credits extend usage past the limit. | Monthly or annual. | Source |
| ChatGPT Enterprise ~150 minimum (reported) |
Not published. Reported $45 – $75, typically ~$60, falling to $40 – $48 above 5,000 seats | Credit-based standard seats include baseline access; eligible additional use draws on credits. Token-based agreements bill in USD under the contract. Eligible credit-based Codex-only seats have no fixed seat fee. | Annual, prepaid (reported). Credit prices and discounts are set per agreement. | Source |
Method
Usage inputs are illustrative dollars per user per month, expressed at Claude Sonnet 5.5 list rates. Equal-token comparisons scale uncached input and output rates; cache reads and writes are excluded, so they are approximate. Non-developer usage is assumed to cost the same on each vendor unless marked included.
The Artificial Analysis benchmark option applies its cost-per-task ratio to developer usage. Benchmark tasks, model settings and results may not match your workload; it is not a predicted saving.
Claude Team and ChatGPT Business show seat fees only; usage past included limits is not modelled. Enterprise defaults assume everyone has a standard chat seat and all entered usage is metered, with no included allowance deducted. Codex-only seats do not provide ChatGPT and must be available in your agreement. The included-usage checkbox assumes non-developer usage fits the allowance; overages are not modelled.
ChatGPT Enterprise defaults to the reported seat-band midpoint and a reported 150 paid-seat minimum (CloudZero); pricing, minimums, commitment and seat eligibility depend on your quote. A seat-only commitment excludes any usage commitment. The 20-seat Claude minimum is for self-serve; sales-assisted requires 50.
Usage defaults:
- Illustrative chat budget: $15.7 a month. Ramp AI Index, median firm AI spend per employee per month, September 2026; a company-wide spending proxy, not measured chat usage. Source
- Developer: $200 a month. Midpoint of Anthropic's $150 – $250 per developer per month across enterprise Claude Code deployments. Source
- Heavy agentic developer: $660 a month. Illustrative $30 × 22 active days; Anthropic says 90% of users stay below $30 per active day, not that this is a monthly percentile. Source
- Capped power user: $1,500 a month. Uber's monthly cap per employee per agentic coding tool, set after it spent its annual AI budget in four months. Source
Excludes regional processing (1.1x on both vendors), long-context surcharges, fast modes and negotiated discounts.
Why the seat is now the small number: what changed in both vendors' enterprise contracts, and what the 12-month commitment locks in.
Read the brief →