Key takeaways
- GitHub's AI usage report, requested from github.com/settings/billing, lists input, output, cache read and cache write tokens for each model over up to 31 days.
- In VS Code the context window popover shows a session's credits and tokens, and in Copilot CLI the /usage command shows per-model token totals.
- Microsoft's consumer and Microsoft 365 Copilot show AI credits or usage limits, not tokens.
To check your GitHub Copilot token usage, download the AI usage report: go to github.com/settings/billing, open AI usage, click Get usage report and GitHub emails you a file with input, output, cache read and cache write tokens for each model. For a single session, hover over a chat response in VS Code or type /usage in Copilot CLI. Microsoft's Copilot app and Microsoft 365 Copilot don't show tokens at all; they count AI credits or usage limits instead.
Token counts matter more for GitHub Copilot than they used to. Since June 1, 2026, Copilot bills in GitHub AI Credits, and the billing docs say every chat or agent interaction is priced by the input tokens sent to the model, the output tokens it writes and the cached tokens it reuses. One credit is $0.01. So the tokens are what you pay for, and the credit balance is the tokens converted into money.
Which Copilot you have
People mean three different products when they say Copilot. Pick yours, because only one of them shows token counts.
| Product | Where it runs | Usage unit | Shows tokens |
|---|---|---|---|
| GitHub Copilot | VS Code, Visual Studio, JetBrains, Copilot CLI, github.com | AI credits | Yes, in the usage report, VS Code and the CLI |
| Microsoft Copilot app and Copilot in Microsoft 365 Personal, Family, Premium | Windows, copilot.microsoft.com, Word, Excel, Outlook | AI credits and usage limits | No |
| Microsoft 365 Copilot Chat at work | Teams, Outlook, the Microsoft 365 app | Copilot credits for metered agents | No, admins see credits |
How to check GitHub Copilot token usage per model
The per-model token breakdown is the most complete record GitHub gives you. GitHub added it on August 11, 2026, and it is available to anyone on a Copilot plan for individuals and to admins of Copilot Business and Enterprise.
- Go to github.com/settings/billing.
- Click AI usage in the sidebar. The page appears only when your account uses AI credits.
- Click Get usage report at the top of the page.
- Choose the report details, then click Email me the report.
- Open the email GitHub sends to your primary address and download the file. The link expires after 24 hours.
According to GitHub's billing reports reference, the AI usage report covers at most 31 days and groups usage by date, model and username. Four columns hold the tokens:
| Column | What it counts |
|---|---|
input |
Tokens sent to the model: your prompt, open files, chat history, tool results |
output |
Tokens the model generated |
cache_read |
Cached tokens read back for the model |
cache_write |
Tokens written to the cache |
Next to them sit quantity, gross_amount, discount_amount and net_amount, which turn the tokens into credits and dollars. You can request only one usage report per account at a time, and GitHub's usage page guide says the detailed data is only in the web interface, not the REST API's usage endpoint.
If you only want totals, the AI usage page itself shows included credits used, any extra usage beyond your plan, a chart of the models you used and a table of credits and cost per model. That page doesn't show tokens, and GitHub's monitoring guide says you generally can't trace credits back to a single prompt or session there.
On Copilot Business or Enterprise, you can't pull the report yourself. Click your profile picture, then Copilot settings, and read Usage this cycle for your credits. Ask an organization owner or billing manager for the token report, since owners see per-user usage only through it.
How to see token usage in VS Code
VS Code shows usage at three levels, according to its guide to optimizing AI credit usage:
- Each reply. Hover over a chat response to see the credits that turn used.
- The whole session. Hover over or click the context window control in the chat input. The popover shows the session's total cost in credits and how many tokens of the context window it has filled.
- The month. Click the Copilot icon in the status bar. The dashboard shows the percentage of your monthly AI credits used, the date the allowance resets, and on Copilot Free your inline suggestions usage.
For raw token counts, open the Agent Debug Logs. Their Summary view gives the session's total token usage, tool calls and duration, and the Cache Explorer view shows prompt cache hit rates and how many input tokens were reused. The model picker also shows each model's cost per token type and a Low, Medium or High cost tier when you hover over it.
How to check token usage in Copilot CLI
Two slash commands cover it, per the Copilot CLI command reference:
/usage session usage metrics, including per-model token totals
/context how much of the context window is used, with a visualization
On token-billed accounts, each model row in /usage also shows its own AI credit consumption. To keep a session from running away, /limits set max-ai-credits 100 sets a soft cap, and copilot -p "your prompt" --max-ai-credits 100 does the same for a one-off run. GitHub's session limit guide says the feature is in public preview, the minimum is 30 credits, and a response already in progress finishes before Copilot stops.
Token usage in other editors
Other editors show credits, not tokens. In Visual Studio, click the Copilot icon in the top right, then Copilot Consumptions. In JetBrains IDEs, click the Copilot icon in the status bar and choose View quota usage. Xcode has the icon in the menu bar, Eclipse in the status bar at the bottom.
For a whole organization, the Copilot usage metrics API reports token usage too. Since October 23, 2026, its reports include a totals_by_client section with requests, sessions, tokens and a model breakdown for each client, including tools such as OpenCode. Enterprise owners, billing managers, organization owners and roles with "View Copilot Metrics" can read it.
GitHub Copilot plans and credits, as of November 17, 2026
| Plan | Price per month | Base credits | Flex allotment | Total credits |
|---|---|---|---|---|
| Free | $0 | Limited chat and agent usage | None | Not stated |
| Pro | $10 | 1,000 | 500 | 1,500 |
| Pro+ | $39 | 3,900 | 3,100 | 7,000 |
| Max | $100 | 10,000 | 10,000 | 20,000 |
Figures are from GitHub's plans page and billing docs. Base credits go first, then the flex allotment, which GitHub says can change as model prices move. The allowance resets at 00:00:00 UTC on the first day of each calendar month, not on your billing date, and unused credits don't carry over. Code completions and next edit suggestions don't use credits on paid plans. Copilot Free gets 2,000 completions a month.
GitHub also lists two discounts for paid users: 10% off model costs with auto model selection in Copilot Chat or the cloud agent, and 30% off with Copilot Fusion through November 30, 2026.
How to check Microsoft Copilot usage
Microsoft doesn't publish token counts for its consumer or work Copilot.
Microsoft 365 Personal, Family, Premium or Pro. Sign in to your Microsoft account and select Services & subscriptions. Microsoft's AI credits support page, last updated in October 2026, says that is where you see your limits and AI credit usage, and that the apps notify you as you approach or reach a limit. Only the subscription owner gets the AI benefits; family members don't share them. The page gives no credit numbers, and chat is listed at the highest usage tier on every plan. For the chat limits in the Copilot app, see our page on the Copilot daily limit.
Microsoft 365 Copilot Chat at work. Users don't see a counter. Admins open the Microsoft 365 admin center, select Reports, then Usage, then Microsoft Copilot, then Credits. The Copilot Credits report shows credits used by people without a Microsoft 365 Copilot license on pay-as-you-go agents, per user, per agent and per billing policy, over the last 7 or 30 days. It flags users above 2,000 and 3,000 credits. Microsoft gives one example: an agent answering a single complex prompt with tenant graph grounding can use 12 credits.
Where the tokens go
If the report shows more tokens than you expected, look at the input column first. VS Code's guide says old chat history gets processed again in a long conversation, so it recommends a new chat for each new task, /compact to summarize a long one, enabling only the tools a task needs, and keeping noisy files out of the agent's reach with files.exclude or search.exclude. The model matters as much: the same tokens cost more credits on a model in the High cost tier.
For the plain credit view without tokens, our earlier guide covers checking GitHub Copilot usage, and Copilot AI credits explains how the token billing works. If you are deciding between tools, see Codex vs Copilot.
FAQ
Does GitHub Copilot show token usage?
Yes. The AI usage report lists input, output, cache read and cache write tokens per model, VS Code shows session token use in the context window popover and the Agent Debug Logs, and Copilot CLI shows per-model token totals with /usage.
How do I check Copilot token usage in VS Code? Hover over a chat response for that turn's credits, hover over the context window control for the session's credits and tokens, or open the Agent Debug Logs Summary view for total tokens.
Do Copilot code completions use tokens or credits? Completions and next edit suggestions aren't billed in AI credits on paid plans. Chat, agent mode, code review, the cloud agent and Copilot CLI are.
Can I see token usage in Microsoft Copilot? No. Microsoft 365 subscribers see AI credit usage under Services & subscriptions in their Microsoft account, and work admins see Copilot credits in the Microsoft 365 admin center. Neither shows tokens.
Frequently asked questions
Does GitHub Copilot show token usage?
Yes. The AI usage report lists input, output, cache read and cache write tokens per model, VS Code shows session token use in the context window popover and the Agent Debug Logs, and Copilot CLI shows per-model token totals with /usage.
How do I check Copilot token usage in VS Code?
Hover over a chat response for that turn's credits, hover over the context window control for the session's credits and tokens, or open the Agent Debug Logs Summary view for total tokens.
Do Copilot code completions use tokens or credits?
Completions and next edit suggestions aren't billed in AI credits on paid plans. Chat, agent mode, code review, the cloud agent and Copilot CLI are.
Can I see token usage in Microsoft Copilot?
No. Microsoft 365 subscribers see AI credit usage under Services & subscriptions in their Microsoft account, and work admins see Copilot credits in the Microsoft 365 admin center. Neither shows tokens.
Sources
- Per-model token breakdown in the usage reportGitHub
- Billing reportsGitHub Docs
- Viewing your usage of metered products and licensesGitHub Docs
- Monitoring your AI usageGitHub Docs
- Usage-based billing for individualsGitHub Docs
- GitHub Copilot plansGitHub
- GitHub Copilot is moving to usage-based billingGitHub
- Copilot CLI command referenceGitHub Docs
- Setting an AI credit session limit in GitHub Copilot CLIGitHub Docs
- Usage metrics API adds usage from tools like OpenCodeGitHub
- Optimize AI credit usage in VS CodeVisual Studio Code
- AI credits and limits for Microsoft 365 subscriptionsMicrosoft Support
- Microsoft Copilot Credits reportMicrosoft Learn


