Vercel adds per-deployment billable duration and CPU minutes to dashboard, CLI, and API

Vercel deployments now report two usage figures that make build costs far easier to attribute: billable duration and CPU minutes. According to the Vercel changelog, the numbers appear in three places at once — the project dashboard, the vc inspect CLI command, and the REST API — giving teams a per-deployment view of how each build contributes to overall usage.

How the metrics are calculated

Vercel defines billable duration as a deployment's build time plus its post-build time, rounded up to the next full minute. To get CPU minutes, Vercel takes that billable duration and multiplies it by the vCPU count of the machine that ran the build.

The changelog includes a worked example that shows how the rounding behaves. A deployment with a build duration of 5 minutes 9 seconds and a post-build duration of 3 minutes 21 seconds has 8 minutes 30 seconds of raw execution time, which is billed as 9 minutes. On a 30-vCPU machine, that deployment consumes 270 CPU minutes.

The rounding matters more as machines get larger. Because CPU minutes scale with the vCPU count, every partial minute that gets rounded up carries the full multiplier — on a 30-vCPU instance, each rounded minute adds 30 CPU minutes to the total.

Viewing the breakdown

In the dashboard, the figures appear alongside each deployment's other details. The REST API exposes the same data, which means teams can pull per-deployment usage into their own tooling — cost dashboards, budget alerts, or per-team chargeback reports — rather than relying on aggregate usage summaries.

For the CLI, Vercel requires version 59.23.1 or later, which can be installed with npm i -g vercel@latest. Once updated, running vc inspect against a deployment prints a usage block listing the build duration, post-build duration, billable duration, and CPU minutes, with the vCPU multiplier shown explicitly so the arithmetic is auditable. In the changelog example, the 270-minute figure is presented as 9 minutes on 30 vCPUs.

Vercel also points readers to its documentation on build machines and CPU minutes for further detail on how the underlying billing model works.

Why it matters

Until now, teams on Vercel have largely had to reason about build usage from aggregate numbers, which makes it hard to answer a simple question: which deployments are actually expensive? Per-deployment billable duration and CPU minutes turn that aggregate into something attributable.

That has practical consequences. A team can spot a single build that dominates usage, evaluate what a change in build machine size does to CPU-minute consumption, or verify that a caching improvement actually reduced billable time rather than just feeling faster. Because the data is available through the REST API, it can also be folded into internal reporting and monitoring automatically instead of being checked by hand in the dashboard.

It is a small change with outsized value for anyone budgeting around Vercel usage: it converts usage from a number that arrives at the end of the billing cycle into a signal attached to the specific deployment that produced it.