Does a stopped Vultr instance still cost money?

Why stopped instances still bill you

When you stop an instance, you tell the hypervisor to power down the guest operating system. That disables network connectivity and halts all your active processes, but your data stays sitting on the SSD storage, and your allocated public IP address remains attached to your account.

According to Vultr's documentation, stopping an instance does not halt billing because the instance continues to reserve critical resources such as CPU, RAM, SSD storage, and IP addresses. Those reserved resources stay allocated exclusively to your instance, so Vultr cannot hand the underlying RAM or disk space to another customer while your server is turned off. Because that slice of the datacenter is held for you, it remains billable.

If you want the charges to stop completely, stopping the server won't do it. You have to destroy the instance. Destroying an instance releases those associated resources back to the pool and permanently deletes its data.

┌──────────────────┐     Stop      ┌──────────────────┐
│ Running Instance │ ────────────> │ Stopped Instance │
└──────────────────┘               └──────────────────┘
        │                                   │
        │ Incurs hourly charges             │ Still incurs hourly charges
        │                                   │ (CPU, RAM, SSD, IP reserved)
        ▼                                   ▼
┌─────────────────────────────────────────────────────┐
│                  Destroy Instance                   │
└─────────────────────────────────────────────────────┘
                          │
                          ▼
            Billing stops permanently;
            resources released & data deleted

If you plan to resume work on that exact machine later today or tomorrow, leaving it stopped makes sense. But if you're done with a workload or putting a project on ice for weeks, leaving a stopped server sitting around is just throwing cash away. In that case, you can take a snapshot of the machine before you destroy it, which preserves your setup for later deployment or migration without paying the full hourly rate of an allocated server. Other hosts handle idle compute similarly, which you can see in our guide on why a powered-off Droplet still bills.

How hourly billing and monthly caps work

Vultr bills every server on an hourly basis. Charges begin the exact moment a server finishes deploying, regardless of whether you keep it powered on or power it down immediately.

To keep monthly invoices predictable, Vultr caps the number of billable hours in any single calendar month. For standard compute workloads, that monthly cap is set at 672 hours. Once an individual instance runs or sits stopped for 672 hours in a calendar month, it stops accumulating additional hourly fees for the rest of that month. If a month has 720 or 744 hours, you aren't charged for those extra hours.

GPU instances follow a slightly different rule. They cap out at 730 hours per month instead of 672.

Invoices generate on the 1st of each month, covering the total accrued charges across all your deployed resources from the previous billing period. If you spin up a server in the middle of the month and destroy it after 10 hours, you pay for 10 hours; if you leave it running or stopped for the full month, you pay up to the cap.

Notice that Bare Metal also caps at 672 hours. You can read more about how other providers handle caps in our overview of which cloud and AI providers let you set a hard spending cap.

How to stop vs destroy an instance

Because the difference between stopping and destroying is the difference between paying and not paying, it helps to be clear on where each action lives.

Stopping an instance

You can stop an instance through the Vultr Console, API, or CLI. If you just need to reboot or temporarily freeze the OS via the web dashboard:

1. Log in to the Vultr Console, navigate to Compute, and click Instances.
2. Click your target Vultr Cloud Compute instance to open its management page.
3. Click the Server Stop icon located in the top-right corner of the management page.
4. Click Stop Server in the confirmation prompt.

At that point, the operating system halts and network access drops. Remember: billing is still running.

Destroying an instance to stop charges

To completely stop incurring charges, you must destroy the instance using the Vultr Console, API, or CLI.

When you destroy it, the hypervisor wipes the local SSD, releases the private and public IP allocations, and removes the compute reservation from your account. The hourly meter terminates at the end of that billing hour. Don't click destroy if you have uncommitted code, unsaved databases, or logs you need, because the deletion is permanent. Take that snapshot first if there's any chance you'll need the environment back. Akamai's platform behaves almost the same way, as detailed in our guide on powered-off Linodes and billing alerts.

How to check pending charges and billing history

Vultr doesn't make you wait until the 1st of the month to see what you've racked up. You can view your current account balance, accumulated charges for the month, and remaining credit whenever you want.

Using the Vultr Console

The web UI provides a clean breakdown of current and past usage:

1. Click Account in the left-hand navigation.
2. Under the BILLING header, click History.
3. At the top of the screen, check your current account balance, charges for the month, and any remaining credit.
4. If you want to review an older invoice, find it in the list and click the ellipsis icon (...) on the right side to download the invoice in PDF or CSV format.
5. If you need a payment receipt for bookkeeping, select the specific payment and click Show Receipt to download it as a PDF.

Using the Vultr CLI

If you spend your day in a terminal, you can pull this data directly with the official CLI without logging into the browser.

The vultr-cli billing command retrieves and displays your Vultr account billing information, including your invoice history, payment details, and current balance. It has two subcommands available:

# Display billing history information, balances, and credits
vultr-cli billing history

# Display invoice information and past line items
vultr-cli billing invoice

Running these commands lets you verify what your stopped and running instances have accrued mid-month before the final invoice drops on the 1st.

Watching your spend with Costbrake

Leaving a high-memory or GPU instance stopped over a long weekend is an easy way to burn twenty or fifty dollars without realizing it. Costbrake watches what you spend across cloud, AI, and SaaS providers, alerting you to sudden spikes, budget thresholds, and month-end forecasts.

Costbrake monitors your Vultr account by reading your pending charges. It pulls your unbilled spend so you don't have to remember to log into the Vultr Console or run terminal commands every Monday morning.

When you set up API access for monitoring, security matters. Vultr API keys inherit the full permissions of the user who generates them, so you shouldn't use an administrative key. The safer approach is creating a dedicated user inside your Vultr account with only the Billing permission enabled, generating the key from that user, and making sure the key's Access Control setting allows all IPv4 and IPv6 addresses so monitoring checks aren't blocked.

Costbrake includes an opt-in killswitch feature called brakes for platforms where pausing or stopping an active resource halts billing, but brakes are not available for Vultr. Because stopping a Vultr server does not suspend billing, and destroying an instance permanently wipes all data, brakes aren't supported for Vultr; Costbrake never deletes user resources.

Instead, Costbrake relies on alerting to keep your Vultr account in check:

• Spike alerts: Triggered when an hour costs more than 3x the usual baseline (calculated from the trailing 7 days) and is at least $5 higher than normal.
• Budget alerts: Sent automatically at 50%, 80%, and 100% of your configured target spend.
• Forecast alerts: Triggered when your month-to-date charges combined with your burn pace over the last 72 hours project that you'll exceed your budget by the end of the month.
• New-charge alerts: Fired if a new billable service appears on your account this month over $1.

The Free tier costs $0 and includes 2 provider connections, hourly checks, 30 days of history, 3 budgets, and email notifications. The Pro plan runs $12 a month (or $120 a year) for 10 connections, 5-minute checks, 12 months of history, Slack, Discord, and webhook alerts. The Team plan is $39 a month for 50 connections and PagerDuty support. You can review full tiers on the Costbrake pricing page.

To connect your account, you can point your coding tool to our agent setup page, or log in and paste your read-only billing key directly through the dashboard.