GitHub Copilot Pricing 2026: What Your Seat Price Actually Includes

Last updated: August 21, 2026

GitHub AI Credits are the metered billing unit for Copilot. One credit equals $0.01 USD. Chat, agent mode, code review, cloud agents, and the CLI consume credits. Code completions do not. GitHub runs two different allowance structures underneath one credit unit: individual plans carry a base amount plus a variable flex allotment, and organization plans carry a standard amount that eligible existing customers have temporarily exceeded since June.

Verdict in one line: The published seat price is not a monthly spending ceiling. Whether September produces an invoice or an interruption depends on three things the seat price does not disclose: your consumption, your paid-usage policy, and your budget configuration.

TIER C — DOCUMENT-FIRST

Future Stack Reviews holds no Copilot Business or Copilot Enterprise seat and observed no live billing dashboard, invoice, or administrator setting for this briefing. Every vendor input came from a GitHub page read on the date stated. All arithmetic performed by Future Stack Reviews is labeled as such.

What happened

GitHub replaced premium request units with GitHub AI Credits on June 1, 2026. The published seat prices did not move. What each seat includes was restructured underneath them.

Existing Copilot Business and Copilot Enterprise customers have received a higher promotional credit amount since June 1. GitHub’s billing documentation gives that period as June 1 to September 1, 2026, and states that included usage returns to the standard amounts afterward. GitHub also states that additional paid usage is enabled by default for organizations and enterprises, and that there is no automatic fallback to a lower-cost model once a budget is exhausted.

Who faces a billing or availability change

  • Highest impact. Existing Business or Enterprise customers already consuming more than the standard allowance. The full reduction converts to billed usage, or the block arrives sooner
  • Sharpest change. Customers consuming between the standard and promotional amounts. They have never hit a limit and may hit one for the first time
  • Finance and platform teams that built forecasts on June through August consumption
  • Any organization that has not confirmed whether paid additional usage is enabled

Who is unlikely to see an immediate effect

  • Organizations consuming below the standard allowance every month
  • Users who rely only on code completions and next edit suggestions, which consume no credits on any paid plan
  • Customers who began after the promotional allocation and were never on the higher amount
  • Individual Pro, Pro+, and Max subscribers, whose plans use a different variable mechanism

What an administrator controls

  • Whether paid additional usage is permitted past the included pool
  • Budgets at user, cost center, organization, and enterprise level
  • Which models are used, since consumption varies by model
  • Whether Copilot code review runs, since it draws on a second meter

What no setting changes

  • Unused credits expire. They do not carry into the next month
  • The pool resets at 00:00:00 UTC on the first day of each calendar month, on a date GitHub states is fixed
  • There is no automatic fallback to a cheaper model when a budget is exhausted
  • On individual plans, GitHub states the flex allotment may change

At a glance

Billing unit1 GitHub AI Credit = $0.01 USD
Model changedJune 1, 2026. Premium request units replaced by AI Credits
Business standard1,900 credits per user per month at $19 per user per month
Enterprise standard3,900 credits per user per month at $39 per user per month
Promotional amounts3,000 and 7,000 credits, for existing customers only
Promotional periodJune 1 to September 1, 2026
Reduction from September 136.7% Business, 44.3% Enterprise (calculated)
Seat pricesUnchanged. Pro $10, Pro+ $39, Max $100, Business $19, Enterprise $39
Paid overage defaultEnabled for organizations and enterprises
RolloverNone. Credits expire at the end of each calendar month
Fallback on exhaustionNone
Second meterCopilot code review also consumes GitHub Actions minutes

Bill or block: the September outcome for Copilot Business

Current use per seatPaid usageWhat changes on September 1
Below 1,900 creditsEither settingNothing changes from this reduction
1,900 to 3,000 creditsEnabledConsumption above 1,900 begins to be billed
1,900 to 3,000 creditsDisabledMetered features stop for the first time
Above 3,000 creditsEnabledAlready paying overage. Billed usage rises by up to 1,100 credits per seat
Above 3,000 creditsDisabledAlready stopping. The stop arrives 1,100 credits per seat earlier
Not measuredNot confirmedNeither the spending nor the availability boundary is known

Enterprise thresholds are 3,900 and 7,000 credits per seat. Credit amounts, the promotional period, and the default paid-usage state are stated by GitHub. The mapping of consumption bands to outcomes is Future Stack Reviews’ reading of those rules and is labeled INFERENCE in the Methodology.

How Copilot usage is metered

GitHub retired premium request units on June 1, 2026 and replaced them with GitHub AI Credits. The change was announced on April 27, 2026 by Chief Product Officer Mario Rodriguez. Token consumption, covering input, output, and cached tokens, is priced per model and converted into credits. One credit is $0.01 USD.

Credits are consumed by Copilot Chat, agent mode, code review, cloud agents, Copilot CLI, Copilot Spaces, Spark, and third-party coding agents. They are not consumed by code completions or next edit suggestions, which GitHub states remain unlimited on every paid plan.

GitHub gave a reason. The announcement stated that a short chat question and a multi-hour autonomous session could cost a user the same amount under the old system, and described the premium request model as no longer sustainable. GitHub also said it had absorbed much of the rising inference cost behind that usage.

One workflow carries two meters. GitHub states that Copilot code review consumes GitHub Actions minutes in addition to AI Credits, billed at the same per-minute rates as other Actions workflows. A team forecasting review volume needs both surfaces, not the credit pool alone.

Sources: GitHub, 27 April 2026 · GitHub Docs, accessed 20 August 2026 · GitHub, accessed 20 August 2026

Two allowance systems, one credit unit

Individual and organization plans use different forms of variable allowance. Reading them as the same mechanism produces the wrong forecast.

On individual plans, GitHub’s plans page separates the advertised total into a base credit amount and a flex allotment, with a footnote stating that flex allowances are not fixed and may change in future.

On Business and Enterprise, the plans page shows base credits equal to the advertised total and a dash in the flex row. GitHub’s organization billing documentation does not describe an ongoing flex allowance for those plans. Instead it publishes a standard monthly amount and a separate, dated promotion.

PlanSeat priceAdvertised or standard creditsVariable component
Copilot Pro$10 / month1,500 ($15)500 flex credits ($5)
Copilot Pro+$39 / month7,000 ($70)3,100 flex credits ($31)
Copilot Max$100 / month20,000 ($200)10,000 flex credits ($100)
Copilot Business$19 / user / month1,900 standard3,000 during the promotion, for existing customers
Copilot Enterprise$39 / user / month3,900 standard7,000 during the promotion, for existing customers

Dollar amounts and the base and flex separation are stated by GitHub. Credit counts for individual plans are converted from those dollar amounts at GitHub’s published rate of $0.01 per credit and are calculated by Future Stack Reviews. Organization credit counts are stated by GitHub.

Two observations follow. On individual plans, between one third and one half of the advertised allowance sits in a layer GitHub describes as changeable. On organization plans the variable component is a promotion with a published end date rather than a standing allotment.

Sources: GitHub, accessed 20 August 2026 · GitHub Docs, accessed 20 August 2026

The September reduction and who it reaches

GitHub’s organization billing documentation publishes both amounts on the same page. It describes the higher allowance as applying to existing Copilot Business and Copilot Enterprise customers for the first three months of usage-based billing, gives that period as June 1 to September 1, 2026, and states that included usage returns to the standard amounts afterward. GitHub’s April announcement describes the same window as June, July, and August.

PlanPromotionalStandard from September 1Change
Copilot Business3,000 credits per user per month1,900 credits per user per month−1,100 credits, −36.7%
Copilot Enterprise7,000 credits per user per month3,900 credits per user per month−3,100 credits, −44.3%

Credit counts are stated by GitHub. Percentages are calculated by Future Stack Reviews.

The affected group is narrower than every account holding a seat. GitHub’s wording is “existing” Business and Enterprise customers, and the promotional period began on June 1. An organization that started later should not assume its pool ever contained the higher amount. The pages read for this briefing do not define the exact eligibility cutoff, and an administrator should confirm the current allowance in the billing dashboard rather than infer it.

Impact then depends on consumption rather than on seat count. An account consuming under 1,900 credits per seat may see no effect from this reduction. An account consuming between 1,900 and 3,000 has been operating inside the promotional allowance but outside the standard one, and is the group most likely to meet a limit it has not met before. An account already consuming more than 3,000 is not insulated. Assuming the same workload, the entire 1,100-credit reduction moves into billed usage or into an earlier stop.

Sources: GitHub Docs, accessed 20 August 2026 · GitHub, 27 April 2026

What 100 seats actually means

Credits are pooled at the billing entity level rather than held per seat. GitHub’s own example uses 100 Copilot Business licenses and gives the resulting shared pool as 190,000 credits. At the promotional amount the same seat count corresponds to 300,000 credits. The reduction is 110,000 credits per month.

That reduction is not an automatic invoice. It becomes one only if consumption crosses the standard pool and paid additional usage is permitted.

100 Business seats, paid usage enabled

Monthly consumptionBilled beforeBilled afterIncrease
150,000 credits$0$0$0
250,000 credits$0$600$600
300,000 credits$0$1,100$1,100
350,000 credits$500$1,600$1,100

Arithmetic: consumption minus the applicable pool, multiplied by $0.01. Promotional pool 300,000 credits, standard pool 190,000 credits. Calculated by Future Stack Reviews from GitHub’s published credit counts and rate.

The table shows a ceiling. Once consumption exceeds the promotional pool, the increase stops growing and holds at 110,000 credits, which is $1,100 per month at the published rate. The maximum exposure from this reduction is therefore bounded per seat, and equals the difference between the two allowances.

A twelve-month figure of $13,200 follows from $1,100 multiplied by twelve. That is a projection, not a current fact. It assumes unchanged seat count, unchanged consumption, unchanged credit pricing, an unchanged standard allowance, and paid usage remaining enabled for a full year. Any one of those changing invalidates it.

Source: GitHub Docs, accessed 20 August 2026

When the pool runs out: bill or block

Three GitHub rules combine to determine the outcome, and none of them is visible in the seat price.

Credits do not roll over. GitHub states that unused credits expire and that the pool resets to the full monthly amount at 00:00:00 UTC on the first day of each calendar month, on a date it describes as fixed and unaffected by when licenses are added, removed, or billed. An organization running below the promotional allowance has no mechanism to bank the difference before September.

Paid additional usage is enabled by default. GitHub states that organizations and enterprises have additional usage on by default, and that an administrator must explicitly disable the AI credits paid usage policy to prevent spending beyond the included credits. That default matters more after September 1 than before it, because the pool behind it has become smaller while the policy has not changed.

There is no automatic fallback. GitHub’s April announcement stated that <a href=”https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/” target=”_blank” rel=”noopener”>”fallback experiences will no longer be available”</a>. Under the previous system, some users who exhausted premium requests could move to a lower-cost model and keep working. The organization billing documentation restates the current position: there is no automatic fallback to a low-cost model when a budget is exhausted.

Taken together, exhaustion produces one of two outcomes. With paid usage permitted, metered work continues and the extra credits are billed. With it disabled, metered features stop until the next monthly reset.

Two mid-cycle behaviors are worth noting because they run in the customer’s favor. Adding licenses during a cycle increases the pool immediately. Removing licenses during a cycle does not shrink the pool until the start of the next billing cycle.

Sources: GitHub Docs, accessed 20 August 2026 · GitHub, 27 April 2026

What a budget does and does not guarantee

GitHub supports budgets at four levels. The documentation read for this briefing describes each as capping metered spending once the pooled credits are exhausted.

User levelLimits what an individual can draw from both the shared pool and additional usage per billing cycle. A $0 user budget blocks that user immediately
Cost centerCaps metered charges for a defined user group after the pool is exhausted
OrganizationCaps metered charges for users whose Copilot seats are billed to that organization
EnterpriseCaps total metered charges across the enterprise after the pool is exhausted

The levels interact. GitHub states that a user can be blocked by an exhausted enterprise spending cap before reaching their own budget limit. An administrator setting only a user-level budget has therefore not defined the enterprise ceiling, and one setting a generous user budget under a tight enterprise cap has not raised the effective limit.

One question this briefing cannot close. The pages read describe budgets as caps but do not, in the text reviewed, distinguish between a control that alerts and a control that stops usage when a threshold is reached. GitHub publishes a separate budget tutorial that Future Stack Reviews did not open for this briefing. An administrator relying on a budget to prevent spend should verify the enforcement behavior of each control in their own account rather than assume that setting a number stops the meter.

Source: GitHub Docs, accessed 20 August 2026

What the seat includes beyond credits

At GitHub’s published rate, the standard included pool has a nominal metered value equal to the seat price. Business is $19 per seat with 1,900 credits, and Enterprise is $39 with 3,900. Multiplying either credit count by $0.01 returns the seat price.

That arithmetic describes how GitHub has structured the retail bundle. It says nothing about GitHub’s infrastructure cost, margin, or internal allocation, and it does not mean the non-metered entitlements come at no charge. What it does mean is that a buyer comparing seat prices across vendors is comparing a figure that, on these plans, is also the size of the metered allowance.

The plans page lists the following on Business and Enterprise, none of them marked as credit-consuming:

Unlimited code completions and next edit suggestionsIncluded on all paid plans. No credit consumption
IP indemnityListed on Business and Enterprise. Not listed on individual plans
Enterprise-grade securityListed on Business and Enterprise. Not listed on individual plans
User management on GitHub.comListed on Business and Enterprise. Not listed on individual plans
Usage metricsListed on Business and Enterprise. Not listed on individual plans
Pooled credits and budget controlsOrganization-wide sharing with administrator spend management
SAML SSO authenticationRequires GitHub Enterprise Cloud. A platform prerequisite, not something the Copilot seat provides on its own

One further difference sits beneath the tier boundary and is not a credit question. GitHub’s model training policy differs between individual and organization plans. Future Stack Reviews is examining that separately.

Source: GitHub, accessed 20 August 2026

What to check before September 1

EligibilityConfirm in the billing dashboard whether the account currently receives 3,000 or 1,900 credits per Business seat. Do not infer it from the signup date.
Burn rateRecalculate August consumption against the standard pool. An account at 65% of 3,000 credits per seat sits at 103% of 1,900.
Paid usage policyConfirm whether the AI credits paid usage policy is enabled. It is on by default for organizations and enterprises.
Budget enforcementVerify for each configured budget whether it alerts or stops usage. Do not assume a number is a hard limit.
Model selectionConsumption varies by model. GitHub states a quick question to a lightweight model may cost a fraction of a credit, while a long agent session on a frontier model across many files costs more.
Code review budgetForecast GitHub Actions minutes alongside credits. Review-heavy teams draw on two meters.
Client versionsGitHub publishes minimum IDE, client, and extension versions. Older versions may show incorrect model pricing, inaccurate usage information, or outdated billing terms.

FAQ

How much does GitHub Copilot cost in 2026?

Copilot Pro is $10 per month, Pro+ is $39, and Max is $100. Business is $19 per user per month and Enterprise is $39. Those prices did not change when GitHub moved to usage-based billing on June 1, 2026. They are not spending ceilings, because metered usage can be billed on top.

Who receives the higher promotional allowance?

GitHub describes it as existing Copilot Business and Copilot Enterprise customers, for the first three months of usage-based billing, given as June 1 to September 1, 2026. The pages reviewed do not define the exact eligibility cutoff, so confirm the current allowance in the billing dashboard.

How much could 100 Business seats pay after September 1?

It depends on consumption. Below 190,000 credits per month, nothing. At 250,000, roughly $600. At 300,000 or above, the increase caps at $1,100 per month, which is the 110,000-credit reduction at the published rate. Those figures assume paid usage is enabled.

Does setting a GitHub budget stop spending automatically?

Not something to assume. The documentation reviewed describes budgets as caps and states that a $0 user budget blocks that user immediately, but it does not distinguish alerting from enforcement for every control. Verify the behavior of each configured budget in your own account.

What happens when AI Credits run out?

With paid additional usage enabled, work continues at the published credit rate and the organization is billed. With it disabled, metered features are blocked until the next cycle. GitHub states there is no automatic fallback to a lower-cost model, which the previous premium request system allowed for some users.

Does Copilot code review use GitHub Actions minutes?

Yes. GitHub states that Copilot code review consumes GitHub Actions minutes in addition to AI Credits, billed at the same per-minute rates as other Actions workflows. A single review workflow therefore draws on two separate billing surfaces.

Does this affect Copilot Pro, Pro+, or Max?

Not this promotion, which GitHub documents for Business and Enterprise. Individual plans carry a base amount plus a flex allotment, and GitHub’s footnote states flex allowances are not fixed and may change. That is a separate mechanism with no published end date.

Methodology, status, and open questions

This is a Tier C briefing. Future Stack Reviews holds no Copilot Business or Copilot Enterprise seat, completed no checkout, observed no invoice or administrator setting, and measured no credit consumption.

Pages read, all on August 20, 2026.

  1. GitHub’s April 27, 2026 announcement of usage-based billing, read in English.
  2. GitHub’s Copilot plans page, individual and business tabs. The business tab was read in the Japanese localization. Prices are stated in USD and are identical across localizations.
  3. GitHub’s usage-based billing documentation for organizations and enterprises, read in the Japanese localization.

Stated by GitHub. Credit unit and rate. Standard and promotional credit counts. Promotional period and its return to standard amounts. Seat prices. Base and flex separation on individual plans, and the footnote that flex is not fixed. The 100-license pooled example of 190,000 credits. No rollover, monthly expiry, and the fixed UTC reset. Default-on paid usage for organizations and enterprises. No automatic fallback. Budget levels and the $0 user budget behavior. Code review consuming Actions minutes. Minimum client versions. The GitHub Enterprise Cloud requirement for SAML SSO.

Calculated by Future Stack Reviews. The reductions of 36.7% and 44.3%. Individual credit counts converted from published dollar amounts at $0.01 per credit. The 110,000-credit difference at 100 seats. All four scenario rows in the 100-seat table. The observation that the standard allowance has a nominal metered value equal to the seat price.

Inferred by Future Stack Reviews. The mapping of consumption bands to bill-or-block outcomes. It follows from GitHub’s stated rules but is not itself published by GitHub.

Projection. The twelve-month figure of $13,200. It assumes five separate conditions hold unchanged for a year.

Open questions.

  • Whether GitHub intends to extend the promotional period. No extension appears in the three pages read.
  • The exact eligibility cutoff for the promotion. GitHub’s wording is “existing customers” and the period began June 1, 2026.
  • Whether every budget control enforces a stop or only alerts. GitHub publishes a separate budget tutorial that was not opened for this briefing.
  • The exact wording of the paid usage policy control inside a live organization account, which requires administrator access.
  • A row on the plans page business comparison labeled as pooled usage and presented as a monthly per-user count rather than a credit amount. It does not reconcile with the credit figures in the billing documentation. It is used in no calculation here and is recorded as unresolved.

Volatility. Pricing pages and billing documentation for AI coding tools change frequently. Confirm current figures on GitHub’s own pages before making a purchase or budget decision.

Sources: GitHub, 27 April 2026 · GitHub, accessed 20 August 2026 · GitHub Docs, accessed 20 August 2026

Verdict

The seat price is the figure buyers compare, and it is not changing. It is also not a monthly ceiling.

Below the standard pool, September 1 passes without effect. Between the standard and promotional amounts, an organization that has never met a limit may meet one. Above the promotional amount, the reduction converts directly into billed usage or into an earlier stop, and that group carries the largest change rather than the smallest.

Which of those two outcomes lands is set by a policy that defaults to billing rather than stopping. An administrator who has not opened that setting has not chosen it.

The decision this briefing supports is therefore not which plan to buy. It is whether, from September, the organization intends to pay for usage above the pool or to have Copilot stop at it. Both are defensible. Neither is the default for a team that has not looked.

Contact

Running Copilot Business or Enterprise and seeing different numbers on your billing dashboard? We want to know. If your budget control behaves differently from what GitHub’s documentation describes, we especially want to know.

[email protected]

Future Stack Reviews corrects published briefings on the record. If a figure here is wrong, tell us and we will say so in the article.

Related FSR briefings

Alternatives are not tiered. Tier B briefings are hands-on tested. Tier C briefings are document-first, with no hands-on testing.

Alternatives

7 GitHub Copilot Alternatives in 2026: What Each One Actually Replaces

Seven tools, read by what each one replaces rather than what each one adds.

Tier C

Cursor Pricing in 2026: Plans, Usage Limits, and What the Public Pages Do Not Quantify

Plans and usage limits, and the parts of the meter the published pages do not quantify.

Tier C

Claude Code Plans and Billing: What Actually Controls a Session

Which plan terms and account settings govern a single session.

Tier C

Claude Pro, Max, Team, and Enterprise: The Control Boundaries Behind the Plan

Four plan names, and the control boundaries that sit behind them.

Future Stack Reviews is an independent publication. This briefing is not sponsored, and no vendor reviewed it before publication. Figures are read from vendor pages on the dates stated and change frequently. Nothing here is legal, financial, tax, or procurement advice. Last updated: August 20, 2026.