GitHub Copilot paid plans include a monthly pool of AI credits, and one credit is worth one US cent. A task costs its tokens at the chosen model's price, converted to credits. To compare with API list prices, estimate the tokens in and out, multiply by the model's per million token rate, and divide by 0.01.
All prices below were fetched on 18 October 2026 from the pages listed under Sources. Prices change, so check them again before you decide anything.
What do the Copilot plans include?
GitHub's individual billing page lists these monthly allowances.
| Plan | Price per month | AI credits per month |
|---|---|---|
| Copilot Pro | $10 | 1,500 (1,000 base plus 500 flex) |
| Copilot Pro+ | $39 | 7,000 (3,900 base plus 3,100 flex) |
| Copilot Max | $100 | 20,000 (10,000 base plus 10,000 flex) |
| Copilot Business | $19 per seat | 1,900 per user |
| Copilot Enterprise | $39 per seat | 3,900 per user |
Base credits are used first, the flex allotment after that. Unused credits do not roll over, and the counter resets at 00:00:00 UTC on the first of each month. GitHub says each interaction consumes input, output and cached tokens, each priced for the model used, and the total becomes credits. A $10 credit is 1,000 credits.
How do I price one task in credits?
Pick a task. Say an agent run reads 80,000 tokens of code and instructions and writes 8,000 tokens back. Use Anthropic's published API list prices per million tokens as one reference, because they are public and easy to check.
| Model | Input rate | Output rate | Task at list price | Credits |
|---|---|---|---|---|
| Claude Haiku 5.5 (prompts up to 100,000 tokens) | $0.10 | $0.50 | $0.012 | about 1.2 |
| Claude Sonnet 5.5 | $2 | $10 | $0.24 | 24 |
| Claude Opus 5.5 | $4 | $20 | $0.48 | 48 |
| Claude Fable 5.1 | $10 | $50 | $1.20 | 120 |
The arithmetic for Sonnet 5.5 is 0.08 million input tokens times $2, plus 0.008 million output tokens times $10, which is $0.16 plus $0.08.
This is list price arithmetic. The Copilot pages we fetched do not state that Copilot charges a model at the same per token rate as the provider's own API, so the credit column shows what the task would cost at API list prices, not a Copilot invoice. Check GitHub's model pricing before you rely on it.
What does that mean for a monthly plan?
Copilot Pro's 1,500 credits equal $15 of usage at one cent each. At 24 credits per Sonnet sized task, that is about 62 tasks a month. At 120 credits per task on the most expensive model in the table, it is about 12. Caching and shorter contexts change the real numbers a lot, so measure a week of your own work.
Which pages disagree?
GitHub's plans page and its individual billing page describe AI credits. A separate GitHub page on premium requests still lists Copilot Pro at 300 premium requests per month, Pro+ at 1,500, and $0.04 for each additional request. It says each model has a multiplier, and it does not print the multiplier table. We could not tell from the fetched text which scheme applies to your account today or when one replaced the other. If your bill uses premium requests, the credit math above does not apply.
What to do
- Open your own Copilot usage page and note which unit it shows, credits or premium requests.
- Log tokens in and out for five real tasks.
- Price them with the formula above for each model you use.
- Compare the monthly total with the plan allowance, then decide whether to stay, move up or cap spending.
- Re-check the published prices each month.
Quick check
How much is one GitHub AI credit?
One credit equals $0.01 US, according to GitHub's billing page. A $10 budget is 1,000 credits.
Is a Copilot task priced the same as the model's API list price?
The pages we fetched do not say. They say usage is priced by model and token type. Treat API list price as a reference point and confirm Copilot's own rates.
Sources
- GitHub Docs: Copilot plans, fetched 18 October 2026
- GitHub Docs: usage-based billing for individuals, fetched 18 October 2026
- GitHub Docs: premium requests, fetched 18 October 2026
- Anthropic API pricing, fetched 18 October 2026
Last reviewed: 18 October 2026.
