Skip to lesson
Lesson 03 / 2010 min readUp to date as of 19 October 2026

Plans and tokens

Subscriptions, team seats and usage

Compare plans and count the seats your team needs. Learn what a token is and why subscriptions and API usage are billed separately.

Reading time is approximate. Allow extra time for the exercises; practice estimates are shown with the workshops.

After this lesson
  • Compare plans and seat billing
  • Distinguish tokens, the context window and usage limits
  • Estimate a pilot budget before buying extra seats
Section being translated

The chrome and course program are available in English. Deep training modules (prompt library, AI glossary, playbooks) are being translated - some content below is still in Polish.

Edition date and feature availability

Material revised: 19 October 2026; The interface and folder guide was revised on 19 October using supplied screenshots and OpenAI’s public plugin repositories; this does not reverify product access. The models in Chat, Work and Codex were checked on 16 October 2026, and the GPT-5.5 retirement date and Word access rules on 14 October 2026; other prices and availability were not fully reverified. Check the documentation and your account before using a feature: access depends on plan, region and administrator settings. Use the examples for practice and review the results yourself.

See how it works

Tokens: capacity and cost are different

  1. Input

    Instructions, conversation and materials use tokens.

  2. Context

    This limits how much information the model can use for a task.

  3. Output

    The response also uses tokens and has its own limit.

  4. API bill

    Input tokens × rate + output tokens × rate. The subscription is billed separately.

A concrete example

A maths example, not a price list: 10,000 input tokens at $2/M and 2,000 output tokens at $10/M cost $0.04 per call.

Does a larger context window mean free API use?

No. Capacity defines the task limit. Billing depends on usage and the terms of the particular product.

Lesson 03 - Corporate subscriptions

How subscriptions work - plans, seats, billing

In shortIn a team plan, a seat gives one person access. Compare price, billing period and limits, then choose a seat count for your pilot. ChatGPT subscriptions and API usage are billed separately.

Follow four steps: choose a plan, set the seat count, invite the team and review administration. Confirm pricing and contract terms before buying.

1

Pick a plan

Compare Business and Enterprise for team size, administration and data requirements. Lesson 04 covers security terms.

2

Buy seats

One seat = access for one person with their own login. Start with a pilot team; check your plan's billing terms before adding seats.

3

Invite the team

Send invitations from the admin panel. After accepting, the employee signs in and selects the company workspace rather than the personal one.

4

Manage in one place

Review team members, roles and billing. Available reports depend on the plan. Before revoking access, preserve materials the company needs under its retention rules.

Seat

The billing unit for a person's access to the team. Seat price and usage limits are separate terms. Check the admin panel and plan agreement before changing the assigned user.

Workspace

An organisation's space with members, projects and settings. A workspace alone does not guarantee that every chat is retained or transferred when someone leaves; check retention and sharing rules.

Monthly vs annual

Compare the price and commitment of both options. You can plan a 2-3 month pilot, then assess annual billing. Check renewal and cancellation terms.

Plan comparison - one table

Indicative prices - check openai.com for the latest. Also check current administration features and data terms.

← Swipe the table →

PlanPriceWho it's forData trains models?Admin panelHow to buy
FreeUSD 0To try what ChatGPT isYes (can be disabled manually)-Sign up at chatgpt.com
Go~USD 8 / moCheap option for individuals (global since 2026)Yes (can be disabled manually)-Card, individual account
PlusUSD 20 / moIndividuals, freelancersYes (can be disabled manually)-Card, individual account
BusinessUSD 25 / user / mo, or USD 20 billed annuallyCompanies and teams of 2+No - by defaultYes: seats, roles, invoicesOnline, company card, 2+ seats
EnterpriseCustomLarge organizations, regulated sectorNo - default + DPAFull: SSO, SCIM, audit logsContact OpenAI sales

← swipe the table to see all columns

Where does the GPT-6 family fit in?

GPT-6 Astra handles the hardest tasks, while GPT-6 Sol and Luna (since 22 Sep 2026) deliver similar quality for less - in Work and Codex on Plus, Pro, Business, Enterprise and Edu; Free and Go get Luna in the desktop app. Choose a model by testing quality, time and usage on your own process. Check access and limits in your account; API billing and limits are separate. See lesson 13 for a worked cost example.

Budget separately for ChatGPT Business seats and API usage. A per-person subscription is not an API token package. API cost depends on the model, input and output tokens, and tools used. In a pilot, measure the cost of one accepted task, not just one call.

What happens when an employee leaves?

The admin revokes access. First decide which materials need to be retained or shared with the team. Removing a user, reassigning a seat and reducing the bill are separate operations; check your plan's rules.

Does everyone in the company need a paid account?

Buy access for the people in your pilot, for example 5-10 people. Match seat count to their tasks and the plan's terms. Review usage after the pilot before expanding access.

How do we migrate from personal Plus to a company Business?

Create a Business workspace and invite employees on work email addresses. Decide whether personal workspaces should remain separate. Check current migration and billing rules before merging workspaces or cancelling Plus. The default no-training rule for Business applies to work in that workspace; it does not automatically change personal chat settings.

Do we get a VAT invoice?

Before buying, check buyer details, tax ID and available documents in the billing panel. Tax treatment depends on the country and buyer's status. Agree Enterprise billing arrangements in the contract.

💡Recommended path for a company: Business, monthly billing, pilot group of 5-10 people - a monthly seat costs more, but you are not tied in for a year. After 2-3 months, once real savings are visible, switch to annual billing (cheaper per seat) and roll out to more departments.Calculate seat costs and compare them with measured pilot benefits before buying.

Lesson 03 - Tokens

Tokens - the currency you pay AI with

In shortA token is a unit of text processed by a model, sometimes part of a word or a character. Tokenisation depends on model and language. Context capacity, output limits and API billing describe different things; product memory is distinct from context.

Enter text to see an illustrative split into units. This demonstrates the concept: check actual tokens and cost with the model's tokenizer or API usage report.

The five-character splitting below only demonstrates a concept. It is not a model tokenizer or cost counter. Check actual usage using the model’s tokenizer or API usage report. The 128,000 limit is illustrative only.

115

Chars

22

Words

~27

Tokens

Text-splitting simulation: colours show illustrative chunks, not actual tokens

Prepareasummaryofthemeetingwiththesalesteamandalistoftasksfornextweek.Tone:concrete,inEnglish.

Capacity simulation (illustrative limit ~128,000 tokens)0.02%

Your text is ≈ 0.04 A4 pages. Estimate - the exact number depends on the model's tokenizer.

What is a token?

The smallest chunk of text the model "thinks" in - usually a fragment of a word. "Automation" isn't one word for AI, it might be "auto-mat-ion" - three tokens. In English 1 word ≈ 1.3 tokens.

Context window

How many tokens the model sees at once - your question + previous chat + uploaded files + its answer. Like a desk: large but finite. What doesn't fit isn't considered.

Input and output tokens

Input = what you send (prompt, files, history). Output = what the model writes. On the API you pay for both separately - a shorter, sharper prompt is real savings.

Why does chat "forget"?

Once the conversation exceeds the context window, the oldest parts drop off. The model doesn't "lose memory" - it just no longer sees the beginning. Long threads: summarize and start fresh.

Approximations — do not use for billing

1 English word
≈ 1.3 tokens
1 A4 page
≈ 500-700 tokens
20-page contract
≈ 12,000 tokens
1000 tokens
≈ 750 English words

How to save tokens (and limits)

  • New topic = new chat. A long thread drags all its history along and eats the limit faster.
  • Paste only the necessary fragment of a document, not the whole thing "just in case".
  • On a long chat ask: "summarize our decisions in 10 bullet points" - then continue in a new chat starting from that summary.
  • In ChatGPT Work split big tasks into smaller steps - each with a clean, small context works faster and cheaper.

Check your understanding

The company bought ChatGPT seats and an app calls the API. How do you budget?

This is a self-assessment. Mark the lesson complete after doing the exercise and reviewing the result.

Do this now

Count how many people actually need a paid plan and pick one option (Go / Plus / Business) for the next month of testing.

How to know you can do it

Keep as evidence: A small pilot budget.

  • Seats match the people in the pilot.
  • Subscription and API are separate budget lines.
  • Assumptions and the source of rates are recorded.

Mark the lesson complete after the exercise and review. Reading progress and quiz answers alone do not confirm practical skill.

Lesson status
In progress — 0% of the page scrolled
Next