
Costs & License Model¶
How you get access at AFRY¶
When you request GitHub Copilot from the Application Kiosk, your section is billed monthly. You receive a Copilot Business licence by default.
To upgrade to a Copilot Enterprise licence (which includes additional features), contact the GitHub admins. (Link to be provided.)
The billing model¶
GitHub Copilot uses GitHub AI Credits for usage-based billing.
What is included in the seat price:
- Seat pricing - Copilot Business is $19/user/month
- Code completions and Next Edit Suggestions are unlimited and do not consume credits
How AI Credits work:
- All other features (chat, agent mode, code review) consume AI Credits
- Credits are spent based on token usage - input, output, and cached tokens - at the published rate for each model
- Unused credits do not roll over - they reset each billing cycle
What is a token?¶
A token is a small chunk of text - roughly ¾ of a word. Every message you send and every response you receive is measured in tokens.
The same task costs very different amounts depending on the model and the type of work:
| Task | Token use | Credit impact |
|---|---|---|
| Inline code completion | Very low | Unlimited - no credits consumed |
| Quick chat question | Low | Minimal |
| Code review of a PR | Medium | Moderate |
| Agent session - implement a feature | High | Significant |
| Long agentic session with large context | Very high | Can be substantial |
Heavier models cost more per token than lighter ones. Choosing the right model for the task matters.
⚠️ The figures below are estimates based on published API rates and are not confirmed.
How far does a Business seat ($19/month = 1,900 AI Credits) go?
One AI Credit = $0.01. The credits are spent at each model's published token rate:
| Model | Blended cost per million tokens | Tokens covered by 1,900 credits |
|---|---|---|
| Claude Sonnet | ~$6.00 | ~3.2M tokens |
| GPT-5 (standard) | ~$3.44 | ~5.5M tokens |
| Claude Opus (heavy) | ~$10.00 | ~1.9M tokens |
What does that mean in practice?
A typical agent-mode session - with repository context, tool calls, reasoning, and edits - consumes roughly 100K–500K tokens.
At that rate, 10–20 agent sessions on a heavy model can exhaust a full month's allowance in a day.
Quick chat questions and code completions are far cheaper; it is sustained agentic work that drives the cost.
For Copilot Business (AFRY's plan)¶
- $19/user/month - includes $19 in monthly AI Credits per user
- Admins can set budget controls at organisation, cost centre, or individual level
- If the pool runs out: code completions still work; chat and agent mode pause until the next billing cycle or until more credits are purchased
Tips to stay within budget¶
- Use Plan mode before Agent mode - planning is cheaper than doing
- Pick the lightest model that can handle the task; save the most capable models for hard problems. The Auto mode will automatically attempt to pick the cheapest model for the task.
- Use skill libraries like Brave Copilot
- Keep context focused - avoid attaching entire codebases when a single file will do
- Monitor your usage via the GitHub Billing Overview page