ChatGPT Business Billing Explained: Seats, Credits and Invoices
How ChatGPT Business billing works: seat prices, prorated seat adds, why removing users doesn't cut the bill, workspace credits, API credits, invoices and VAT.
Gemini business pricing confuses people because "Gemini" is three separate purchases with three separate bills. There's Gemini inside Google Workspace, paid per user with your Workspace licence. There's Gemini Enterprise, Google's agent app, paid per seat. And there's the Gemini API on Google Cloud, paid per million tokens. This guide takes each in turn: what it costs, when you're charged, the traps, and the South African payment and VAT details.
If you're still deciding whether Gemini is the right platform at all, start with our Gemini platform overview. For the same guide on Anthropic's side, see Claude Team billing explained.
| What you buy | Priced | Billed through |
|---|---|---|
| Gemini in Google Workspace | Per user per month, inside the Workspace plan | Your Workspace subscription |
| Gemini Enterprise (agent app) | Per seat per month, or pay-as-you-go | Google Cloud |
| Gemini API / Agent Platform | Per million tokens in and out | A Google Cloud billing account |
Most mid-sized businesses start with the first and only reach the third when they automate a process. Knowing which line you're on avoids the classic mistake of buying a second AI licence for capability you already pay for.
Gemini is now included in every Workspace Business plan, at different depths. These are Google's published US-dollar prices per user per month:
| Plan | Annual plan | Flexible plan | Gemini included |
|---|---|---|---|
| Business Starter | USD 7 | USD 8.40 | Gemini in Gmail |
| Business Standard | USD 14 | USD 16.80 | Plus Docs, Sheets, Slides, Meet and Chat |
| Business Plus | USD 22 | USD 26.40 | As Standard, with stronger security and archiving |
Business plans are capped at 300 users. Higher tiers can add an AI Expanded Access add-on for higher limits, priced by local currency and term. (Source, Source)
Before buying anything, check which tier you're on. Because Google folded Gemini into the plans, plenty of businesses already pay for it and haven't switched it on.
This is where Workspace bills surprise finance teams.
It's the same trap as on most seat-based AI plans: someone leaves, the licence frees up, and the bill doesn't move. Go annual for the stable core of the team and keep fast-changing groups, such as contractors and seasonal staff, in mind when you set the count.
Gemini Enterprise is Google's agent app: one place to search company data and run agents across Workspace, Microsoft 365 and other systems. It's billed per seat per month and needs an active user licence.
Since August 2026 you can also mix seats with a pay-as-you-go edition, billed at standard model API rates with no upfront commitment. Quotas pool across users in a project, admins decide whether overage is allowed, and Flexible Savings Plans take 10% off a one-year or 20% off a three-year monthly spend commitment. (Source)
Anything you build on Gemini Enterprise Agent Platform (the evolution of Vertex AI) or the Gemini API is pay-as-you-go per million tokens, on a Google Cloud billing account. Prices are listed in US dollars. Here are the two models we use most, on Google's standard tier and global endpoint:
| Model | Input | Cached input | Output |
|---|---|---|---|
| Gemini 4 Argon (from 1 Dec 2026) | USD 4.00 | USD 0.20 | USD 20.00 |
| Gemini 3.8 Flash (until 31 Dec 2026) | USD 0.75 | USD 0.075 | USD 3.75 |
Figures are per million tokens. (Source)
Five details that change the bill:
Google Cloud spend caps became generally available on 30 October 2026, and they're the control we switch on first. For AI services a cap triggers within minutes using estimated costs, resets automatically each month, and alerts at 50%, 80% and 100% by default. One exception: Provisioned Throughput is never halted by a cap. (Source)
Budgets and alerts on their own only send email. A cap stops spending. On a new agent, set both.
The licence is the cheap part. The return comes from the processes Gemini takes over: inference running inside BigQuery, documents classified on arrival, agents grounded in your own data. Our Gemini implementation service covers licence planning and spend controls through to production agents on Google Cloud, and because we're model-agnostic we'll tell you when a workload belongs on Claude, GPT or Copilot instead.
Want to know what Gemini should cost for your team? Book a Free AI Assessment and we'll price it against your actual tier and volumes.
Next step
Our free AI assessment is a scoped conversation about your systems, your constraints and what is actually worth automating — not a product demo. You leave with a plan you can act on, whether or not you go on to work with us.
Get a Free AI AssessmentHow ChatGPT Business billing works: seat prices, prorated seat adds, why removing users doesn't cut the bill, workspace credits, API credits, invoices and VAT.
Claude Code is billed differently depending on how you sign in. Subscription, Team seats or API key — what each costs, the API key trap, and how to cap team spend.
Claude Team, Max and Ultra plans include up to $500 a month in API credits that don't appear in claude.ai. Who qualifies, how to claim them, and what they cover.