GitHub Copilot AI Credits: Plans, Overage and Team Budgets
GitHub Copilot Pro starts at $10 a month, but agentic tasks and deep conversational chats can quickly draw down usage allowances. Choosing the right tier comes down to knowing how your daily workflow consumes GitHub AI Credits and putting firm spend caps in place before unleashing autonomous agents. Standard inline code completions stay completely unmetered on all paid plans.
Pricing and billing rules verified on 3 November 2026 against official GitHub documentation (USD, excluding tax). GitHub meters usage at 1 AI Credit = $0.01 USD. Autocomplete suggestions do not use credits.
Get The NotCheapAI Brief →BILLING CHECK
Before you pay: Copilot team budgets
Copilot Business and Enterprise. These organization controls and defaults do not describe personal Pro, Pro+ or Max accounts.
These billing rules checked .
- Billing unit
- 1 AI Credit equals $0.01 USD. Chat and agent work consume credits; code completions and next-edit suggestions remain unmetered on paid plans. Team billing (opens in a new tab).
- Pool and expiry
- Included credits are pooled at the billing entity. Unused included credits do not roll over; the pool resets on the first of each calendar month at 00:00 UTC. Credit pooling (opens in a new tab).
- Additional usage
- Paid additional usage is enabled by default for organizations and enterprises. To block spending beyond included credits, an administrator must disable AI credits paid usage in AI Controls. Additional-usage policy (opens in a new tab).
- Alert versus hard stop
- Enterprise and cost-center spending limits only notify by default. Enable “Stop usage when budget limit is reached” on each limit to block metered usage at that limit. Spending-limit setup (opens in a new tab).
- Per-user limits
- User-level budgets enforce a hard stop automatically, including use of the shared pool. A $0 user budget blocks that user immediately, even if pool credits remain. User-level budgets (opens in a new tab).
- Seat charges
- Organizations and enterprises pay for assigned seats each monthly billing cycle. Controlling metered usage does not remove a seat or cancel its subscription. License billing (opens in a new tab).
How GitHub AI Credits work and what is unmetered
GitHub uses usage-based billing measured in GitHub AI Credits (1 AI Credit = $0.01 USD). Under official Copilot billing rules, your activity separates into two distinct tracks:
- Unmetered on all paid plans (0 AI Credits): Real-time inline code completions and next-edit suggestions inside your IDE. (Copilot Free caps completions at 2,000 per month).
- Metered in AI Credits: Copilot Chat, Agent Mode in the IDE, Copilot CLI, agentic Code Review, Copilot Spaces, Spark, third-party coding agents, and Copilot cloud agent (which also draws from your account's monthly GitHub Actions minutes allowance).
Instead of charging a flat fee per prompt, each metered interaction consumes input tokens, cached input tokens, and output tokens at the selected model's rate, converting the dollar total into AI Credits at $0.01 per credit. Selecting Auto in the model picker routes tasks automatically and qualifies for optimized rates.
Individual and team plans compared
| Plan | Monthly fee | Included monthly AI Credits | Key billing & access rules |
|---|---|---|---|
| Free | $0 | Small monthly allowance | 2,000 completions/mo; Auto model selection only; excludes third-party agents |
| Pro | $10 / user | 1,500 ($15.00 value: 1,000 Base + 500 Flex) | Unlimited completions; standard model picker; cloud agent & CLI included |
| Pro+ | $39 / user | 7,000 ($70.00 value: 3,900 Base + 3,100 Flex) | Adds premium models (e.g., Claude Opus) and OpenAI Codex VS Code sign-in |
| Max | $100 / user | 20,000 ($200.00 value: 10,000 Base + 10,000 Flex) | Priority access to new models/features; high-volume agent capacity |
| Business | $19 / seat | 1,900 / seat ($19.00 value, pooled across org) | No seat minimum; shared credit pool; paid overage enabled by default |
| Enterprise | $39 / seat | 3,900 / seat ($39.00 value, pooled across org) | Requires GitHub Enterprise Cloud; priority models & enterprise governance |
Individual paid tiers split their allowance into Base credits (matching sticker price) and Flex allotment (bonus credits). Unused included AI Credits expire at 00:00:00 UTC on the 1st of each calendar month on all plans. Mid-cycle changes are prorated under GitHub’s billing rules; inspect the displayed charge and remaining allowance before confirming an upgrade.
Selected model token rates
| Model | Standard Input / 1M | Cached Input / 1M | Cache Write / 1M | Standard Output / 1M |
|---|---|---|---|---|
| GPT-5.4 nano | $0.20 (20 credits) | $0.02 (2 credits) | Not applicable | $1.25 (125 credits) |
| GPT-5.3-Codex | $1.75 (175 credits) | $0.175 (17.5 credits) | Not applicable | $14.00 (1,400 credits) |
| Claude Sonnet 4.6 | $3.00 (300 credits) | $0.30 (30 credits) | $3.75 (375 credits) | $15.00 (1,500 credits) |
| Claude Opus 5 | $5.00 (500 credits) | $0.50 (50 credits) | $6.25 (625 credits) | $25.00 (2,500 credits) |
These are selected Copilot rates, not a list of the newest or cheapest models. Check GitHub’s model table for newer alternatives and access restrictions. Count cache creation separately from cache reads where the model charges for it. The example below selects GPT-5.3-Codex manually, so it does not assume an Auto discount.
Solo break-even: Pro vs. Pro+
Suppose a solo developer works 20 days a month using unlimited autocomplete (0 credits) and runs IDE chat plus local agent refactors using GPT-5.3-Codex. Over the month, metered sessions process 6M uncached input tokens ($10.50 / 1,050 credits), 20M cached input tokens ($3.50 / 350 credits), and 1.2M output tokens ($16.80 / 1,680 credits), totaling 3,080 AI Credits ($30.80 gross usage).
1. Option A — Copilot Pro ($10/mo) + Additional Usage:
Included Pro credits = 1,500 AI Credits ($15.00 value)
Billed overage credits = 3,080 - 1,500 = 1,580 AI Credits
Overage charge ($0.01/credit)= 1,580 * $0.01 = $15.80
Total cash paid this month = $10.00 + $15.80 = $25.80
2. Option B — Upgrade to Copilot Pro+ ($39/mo):
Included Pro+ credits = 7,000 AI Credits ($70.00 value)
Credits consumed = 3,080 AI Credits
Unused at month end, if no further usage = 3,920 AI Credits
Total cash paid this month = $39.00The solo decision rule: Stepping up from Pro ($10) to Pro+ ($39) adds $29.00 in fixed monthly cost. At $0.01 per overage credit, adding 2,900 credits to Pro's 1,500 included credits sets the exact break-even point at 4,400 AI Credits/month ($44.00 gross usage). If you consume between 1,500 and 4,399 credits on standard models, staying on Pro and paying overage saves money. At 4,400 credits the two bills are equal; above that point Pro+ costs less, assuming the same model and features are available. Beyond 13,100 credits/month, upgrading to Max ($100) becomes cheaper than paying overage on Pro+.
Team pooling and budget controls
On Copilot Business ($19/seat/month, no minimum seats required), credit allowances pool across the organization. For a 5-developer team:
1. Fixed license cost and shared pool:
5 seats * $19.00/seat = $95.00/month
Shared org credit pool = 5 * 1,900 = 9,500 AI Credits ($95.00 value)
2. Consumption offset:
3 light users (3 * 800) = 2,400 AI Credits
2 heavy agent devs (2 * 4,500)= 9,000 AI Credits
Total team consumption = 11,400 AI Credits
Net overage beyond pool = 11,400 - 9,500 = 1,900 AI Credits
3. Total monthly organization bill:
Overage charge (1,900 * $0.01)= $19.00
Total cash paid = $95.00 + $19.00 = $114.00/month ($22.80/dev average)Because paid overage is enabled by default on organization plans, administrators should configure a User-Level Budget (ULB)—such as $50.00 (5,000 credits) for each heavy user in this scenario. A ULB counts both included and extra consumption: a $25 cap would stop a developer before the planned 4,500 credits ($45 of usage), even if the organization still had credits. Set an organization overage limit as well; user limits alone do not cap the combined bill.
For comparing editors or estimating editor stacks, see our Cursor vs. GitHub Copilot cost comparison, GitHub Copilot cost calculator, and hidden AI subscription cost audit.
Common questions
Are Copilot AI Credits the same as code completions?
No. GitHub code completions and next-edit suggestions do not use credits on paid plans. Chat, agent mode, code review, CLI, and cloud agents draw from the credit allowance.
Do Copilot AI Credits cover external OpenAI or Anthropic API keys?
No. Copilot AI Credits only meter interactions inside Copilot surfaces. Copilot CLI supports Bring-Your-Own-Key (BYOK) for local or external models, billed directly by your external provider.
Why did my Copilot cloud agent task use both AI Credits and Actions minutes?
Token reasoning draws from your Copilot AI Credits allowance, while the underlying runner environment uses your monthly GitHub Actions compute minutes.
Does upgrading from Pro to Pro+ mid-month double-bill me?
GitHub prorates plan changes. The $29 difference is the full-month fee difference, not necessarily the charge on a mid-month invoice. Check the upgrade preview for the actual charge and credit allocation.
Sources: GitHub Copilot Billing Documentation, GitHub Copilot Plans, and GitHub Copilot Organization Billing, and Copilot model rates. Report a correction with current official documentation.