DigitalOcean Basic 4 GiB: Per-Second Billing, Monthly Cap

Plan details (official data)

Inclusions and features (official data)

• Flat monthly price (no intro discount)

Prices were checked on DigitalOcean's official pricing page on Oct 15, 2026. The renewal price listed there is the same as the sign-up price (no intro discount). Prices change, so confirm on the official site before you buy.

Compare plans in this category (official data)

Swipe the table to see more →

All values in this table come from each host's official pricing page. This plan was checked on Oct 15, 2026; every compared plan was checked within the last 60 days.

What the official data lists for this plan

The plan is a self-managed Cloud VPS. Its resources, in the wording of the pricing page, are 2 vCPU, 4 GiB of RAM, 80 GiB SSD and 4,000 GiB of transfer. The single feature listed is "Flat monthly price (no intro discount)", and the renewal price is listed as the same as the intro price. The term is 1 month.

Several things a developer usually wants to know before sizing a server are not in the official data: the CPU model or generation, data center regions, available operating system images, IPv6 availability, provider-side backup or snapshot options, and the mechanics of resizing to another plan. None of these should be assumed; confirm each one with DigitalOcean before signing up if your app depends on it. The rest of this page refers back to this list rather than repeating it.

How per-second billing up to a monthly cap works

The billing field reads "USD per month; billed per second up to the monthly price (capped at 672 hours a month)". In plain terms, charges build up second by second, and once they reach the monthly price shown in the price table they stop climbing for that month. The feature "Flat monthly price (no intro discount)" describes that monthly figure, but because billing is per second, treat the number in the table as the most you pay in a month rather than as an identical charge that lands every month regardless of use.

The 1 month term is the cycle over which that cap is measured. The official data does not say what, if anything, pauses or lowers the per-second charges, so do not plan around powering a server off, deleting it, or leaving it idle until DigitalOcean has confirmed how those cases are billed. Since the renewal price is listed as the same as the intro price, the monthly cap you see in the table is the one that applies in later months as well.

Matching 2 vCPU and 4 GiB of RAM to a web app

A useful way to size memory is to add up the resident footprint of each process that will run on the box: the application runtime and its worker processes, a database if it lives on the same server, a reverse proxy such as Nginx, any in-memory cache, and the operating system itself. A web app that keeps the database on the same server and runs a handful of workers tends to be the kind of workload where 4 GiB is considered, while a stateless app that talks to an external database may fit in less. Measure your app's memory use under realistic traffic on a development machine before committing to a size.

The 2 vCPU figure matters for request concurrency and for anything CPU-bound that runs alongside the app, such as asset builds, image processing, scheduled jobs or background queues. The data does not describe the CPU type, as noted above, so compare on the vCPU count alone and avoid assumptions about per-core speed. If your framework spawns one worker per core, 2 vCPU also sets a natural starting point for your worker count.

Sizing 80 GiB SSD storage and 4,000 GiB of transfer

The 80 GiB SSD figure has to cover the operating system, your application code and dependencies, the database files if the database is local, logs, uploaded media, and any local copies of backups you keep on the server before shipping them elsewhere. Logs and uploads are the two that grow quietly, so decide early whether uploads will live on the server or in object storage, and set up log rotation on day one.

The 4,000 GiB of transfer is the monthly transfer listed for the plan. To estimate your usage, multiply average response size by expected requests, then add the size of assets you serve directly rather than through a CDN, and add outbound backup traffic. The official data does not say what happens if transfer exceeds that figure, so confirm overage handling with DigitalOcean if your app serves large downloads or video.

Resizing later: what the data does and does not say

The alternatives list includes DigitalOcean Basic 2 GiB / 2 vCPU, which shows that this plan belongs to a family of sizes, and the per-second billing model is the kind of model developers often use to start on one size and adjust once real traffic arrives. However, the official data does not describe how a resize is performed, whether it requires downtime, whether storage can be changed independently of RAM and vCPU, or whether moving to a smaller plan is possible once the disk has grown. Those points are on the unconfirmed list above and are worth checking before you build a scaling plan around them.

A practical approach is to pick the size that fits your measured needs with headroom for memory spikes, run monitoring for memory, CPU steal, disk usage and outbound transfer from the first day, and decide on a resize only once you have a few weeks of data. Keep your deployment reproducible so that, if resizing turns out to involve rebuilding on a new server, moving is a matter of running your provisioning scripts and updating DNS.

What to check before signing up

Read the price table for the monthly cap, remembering that charges accrue per second up to that figure and that the renewal price is listed as the same as the intro price. Confirm with DigitalOcean how charges behave when a server is powered off or deleted, because the data does not cover that. Run through the list of unconfirmed items above, especially regions, backups and resize mechanics, and get answers from the provider before you pay.

Because this is a self-managed server, you are responsible for the operating system, security updates, the web server, the firewall, TLS certificates and backups. Budget time for that administration, or plan to automate it with configuration management, before choosing a self-managed plan over a managed one.

How it compares with DigitalOcean Basic 2 GiB / 2 vCPU and the Vultr plans

DigitalOcean Basic 2 GiB / 2 vCPU lists the same 2 vCPU with 2 GiB of RAM, 60 GiB SSD and 3,000 GiB of transfer, the same "Flat monthly price (no intro discount)" feature, the same per-second billing up to the monthly price capped at 672 hours a month, a 1 month term and a renewal price the same as the intro price. It fits an app whose measured memory use is comfortably below half of what this plan offers, for example a stateless service with an external database, while the 4 GiB plan fits an app that also hosts its database or runs several workers.

Vultr Cloud Compute 4 GB / 2 vCPU lists "Regular Performance (previous-gen Intel, regular SSD)", 2 vCPU, 4 GB of RAM, 80 GB SSD and 3 TB of transfer. Vultr High Frequency 4 GB / 2 vCPU lists "High Frequency (3GHz+ Intel Xeon, NVMe SSD)", 2 vCPU, 4 GB of RAM, 128 GB NVMe SSD and 3 TB. Vultr High Performance AMD 4 GB / 2 vCPU lists "High Performance (AMD EPYC, NVMe SSD)", 2 vCPU, 4 GB of RAM, 100 GB NVMe SSD and 5 TB. All three carry a "flat monthly price (no intro discount)" feature, a 1 month term, a renewal price the same as the intro price, and are billed in USD per month, billed hourly up to the monthly price.

Two things stand out when comparing. First, the Vultr data names the CPU and disk type while the DigitalOcean data does not, so a developer who needs a documented CPU generation or NVMe storage has that information only on the Vultr side. Second, the units differ: DigitalOcean lists GiB and the Vultr plans list GB and TB, so do not convert or rank the storage and transfer figures across providers; compare them on each provider's own page. Among the Vultr plans, High Frequency lists the largest NVMe storage at 128 GB NVMe SSD, and High Performance AMD lists the largest transfer at 5 TB, so a media-heavy app leans toward the latter and a database-heavy app toward the former. DigitalOcean's per-second billing versus Vultr's hourly billing is a difference in how the charges accrue up to the cap, and which one matters depends on how often you create and remove short-lived servers.