GitHub AI Credits Calculator
Since June 1, 2026 every Copilot plan bills in AI credits: 1 credit = $0.01, consumed by tokens at per-model rates. Estimate what your real usage costs and whether your plan allowance covers it. Everything runs in your browser.
Runs entirely in your browser. Nothing you enter leaves this page.
Rates verified against GitHub's official pricing table on 2026-07-14.
GitHub AI credits are Copilot's usage-based billing unit since June 1, 2026. One credit costs $0.01, and each request spends credits based on the model's per-token rates. Included per month: Copilot Pro 1,500, Pro+ 7,000, Max 20,000, Business 1,900 per user, Enterprise 3,900 per user. Code completions stay free. Rates verified against GitHub's published rate sheet on 2026-07-14.
New to AI credits? See how GitHub AI credits work, or browse credit cost by model.
Your usage
Estimated burn
Effective monthly cost: $10.00 plan + $4.99 overage = $14.99/mo
- Credits per request
- 3.8
- $0.04
- Credits per month
- 1,999
- $19.99
This usage exceeds the included allowance. Projected additional usage: $4.99/mo on top of the $10.00/mo subscription. You would need a budget of at least that amount at github.com/settings/billing/budgets, or a higher plan.
Included credits run out after about 15 working days at this pace.
Note: Anthropic models also bill cache writes ($3.75/1M tokens), which this estimate does not include. Real costs run slightly higher on cache-heavy sessions.
Which plan should you be on?
At about 1,999 credits a month, Copilot Pro is already the cheapest option for you at $14.99/mo. No need to upgrade.
Copilot Pro
Best value$14.99/mo
$10.00 plan + $4.99 overage
1,500 credits included
Copilot Pro+
$39.00/mo
$39.00 plan
7,000 credits included
Copilot Max
$100.00/mo
$100.00 plan
20,000 credits included
Overage is $0.01 per credit on every plan, so the cheapest plan is a fixed function of your monthly burn: Copilot Pro is cheapest below 4,400 credits per month, Pro+ from 4,400 to 13,100, and Max above 13,100. Often the right move is not a bigger plan but a small monthly budget.
Included AI credits by plan
| Plan | Price | Base credits | Flex allotment | Total / month |
|---|---|---|---|---|
| Copilot Pro | $10.00/mo | 1,000 | 500 | 1,500 |
| Copilot Pro+ | $39.00/mo | 3,900 | 3,100 | 7,000 |
| Copilot Max | $100.00/mo | 10,000 | 10,000 | 20,000 |
| Copilot Business | $19.00/seat/mo | - | - | 1,900 |
| Copilot Enterprise | $39.00/seat/mo | - | - | 3,900 |
Business and Enterprise credits pool at the billing entity. Existing org customers get promotional amounts (3,000 / 7,000 per user) until September 1, 2026. Copilot Free and Student include an AI-credit allowance whose exact number GitHub has not published. Code completions stay unlimited on paid plans.
Official per-model token rates
USD per 1 million tokens, default tier, as published by GitHub. These are the rates behind every credit your bill shows.
| Model | Vendor | Input | Cached | Output |
|---|---|---|---|---|
| GPT-5.6 Luna | OpenAI | $1.00 | $0.100 | $6.00 |
| GPT-5.6 Sol | OpenAI | $5.00 | $0.500 | $30.00 |
| GPT-5.6 Terra | OpenAI | $2.50 | $0.250 | $15.00 |
| GPT-5.5 | OpenAI | $5.00 | $0.500 | $30.00 |
| GPT-5.4 | OpenAI | $2.50 | $0.250 | $15.00 |
| GPT-5.4 mini | OpenAI | $0.75 | $0.075 | $4.50 |
| GPT-5.4 nano | OpenAI | $0.20 | $0.020 | $1.25 |
| GPT-5.3-Codex | OpenAI | $1.75 | $0.175 | $14.00 |
| GPT-5 mini | OpenAI | $0.25 | $0.025 | $2.00 |
| Claude Fable 5 | Anthropic | $10.00 | $1.000 | $50.00 |
| Claude Opus 4.8 | Anthropic | $5.00 | $0.500 | $25.00 |
| Claude Opus 4.8 fast modepreview | Anthropic | $10.00 | $1.000 | $50.00 |
| Claude Sonnet 5 | Anthropic | $2.00 | $0.200 | $10.00 |
| Claude Sonnet 4.6 | Anthropic | $3.00 | $0.300 | $15.00 |
| Claude Haiku 4.5 | Anthropic | $1.00 | $0.100 | $5.00 |
| Gemini 3.5 Flash | $1.50 | $0.150 | $9.00 | |
| Gemini 3.1 Propreview | $2.00 | $0.200 | $12.00 | |
| Gemini 3 Flashpreview | $0.50 | $0.050 | $3.00 | |
| Gemini 2.5 Pro | $1.25 | $0.125 | $10.00 | |
| MAI-Code-1-Flash | Microsoft | $0.75 | $0.075 | $4.50 |
| Raptor mini | GitHub | $0.25 | $0.025 | $2.00 |
| Kimi K2.7 Code | Moonshot AI | $0.95 | $0.190 | $4.00 |
Anthropic models additionally bill cache writes (1.25x input). Several OpenAI and Gemini models have long-context tiers above 272K / 200K input tokens. Copilot code review bills credits plus GitHub Actions minutes, and its model is auto-selected. Prior-generation models stay selectable in Copilot at their family rates; the full list is on the bill migrator.
Frequently asked questions
What is a GitHub AI credit?
A GitHub AI credit is the usage-based billing unit for Copilot since June 1, 2026. 1 AI credit = $0.01 USD. Each Copilot interaction consumes input, output, and cached tokens, priced at the per-model rates GitHub publishes, and the resulting dollar amount is converted to credits.
How many AI credits does my Copilot plan include?
Copilot Pro ($10/mo) includes 1,500 credits per month (1,000 base + 500 flex). Copilot Pro+ ($39/mo) includes 7,000 (3,900 base + 3,100 flex). Copilot Max ($100/mo) includes 20,000 (10,000 + 10,000). Copilot Business includes 1,900 per user and Copilot Enterprise 3,900 per user, pooled at the billing entity, with promotional amounts of 3,000 and 7,000 until September 1, 2026.
Is Copilot Pro+ worth it?
It depends on your monthly credit burn. Every plan bills overage at the same $0.01 per credit, so Copilot Pro plus a small budget is cheaper until about 4,400 credits per month, where Pro+ overtakes it. Pro+ then wins until about 13,100 credits per month, above which Copilot Max is cheapest. Enter your usage above to see your own break-even.
Which Copilot model uses the fewest credits per request?
Among current models the lightweight tiers cost the least per token: Gemini 3 Flash ($0.50 in / $3 out per 1M tokens), GPT-5 mini and Raptor mini ($0.25 in / $2 out), and GPT-5.4 nano ($0.20 in / $1.25 out) are the cheapest. Powerful models such as Claude Fable 5 or GPT-5.5 cost roughly 10 to 20 times more per token. Auto model selection also gives paid plans a 10 percent discount.
Are code completions billed in AI credits?
No. Code completions and next edit suggestions are not billed in AI credits and remain unlimited on all paid Copilot plans.
What happens when I run out of included credits?
On individual plans you either set a budget in USD and pay for additional usage at the same rates ($10 covers 1,000 credits), or wait for the monthly reset. Organizations choose per policy whether usage continues pay-as-you-go or is blocked until the next cycle.
Do unused AI credits roll over to next month?
No. Included AI credits do not carry over. Unused credits are forfeited and the monthly allowance resets at 00:00 UTC on the first day of each month.
What happens to Business and Enterprise promotional credits on September 1, 2026?
The enhanced promotional amounts (3,000 credits per user on Business, 7,000 on Enterprise) end on September 1, 2026. After that, included credits are the standard 1,900 per Business user and 3,900 per Enterprise user, still pooled at the billing entity. Teams near their promo allowance today can expect a higher effective bill in September.
I have an annual Copilot plan. Does this apply to me?
Annual Pro and Pro+ subscribers who stayed on legacy premium-request billing keep the PRU model until their annual plan ends, then the plan auto-downgrades to Copilot Free unless changed. Monthly plans were all moved to AI credits on June 1, 2026.
Does auto model selection really save money?
Paid plans get a 10 percent discount on model costs when using auto model selection in Copilot Chat, Copilot CLI, or the coding agent. The calculator applies this discount when the toggle is on.
Sources
Prices last verified 2026-07-14, from each provider's official page.
- GitHub Docs: Models and pricing for GitHub Copilot
- GitHub Docs: Usage-based billing for individuals
- GitHub Docs: Usage-based billing for organizations and enterprises
- GitHub Blog: Copilot is moving to usage-based billing
This is an independent community tool, not affiliated with GitHub. Estimates only; your bill is computed by GitHub. GitHub assembles each request's context internally and does not publish tokens per interaction, so per-request costs here are your estimates, not measurements. Rates change as models change, so check the sources for current numbers.
