Node.js on Vercel
Vercel documents a generic path that covers Node.js, rather than a dedicated guide.
JavaScript runtime used to run servers and build tooling outside the browser.
What Node.js needs.
Every host claims to run Node.js, so the claim on its own decides nothing. These are the things Node.js actually requires from the platform underneath it, and what Vercel publishes about each. Vercel documents 2 of the 4.
Persistent processes
Not documented
A long-running process is the point: WebSocket connections, in-memory caches and job queues all die with it. A host that only runs functions cannot keep one alive.
Max request duration
Hobby: 300s default and maximum. Pro and Enterprise: 300s default, 800s maximum, and 1800s extended maximum
Anything the process does inside a request — a large upload, a slow query, a streamed response — is bounded by this ceiling.
Sources (1) →Sources ↓
- Functions limitations — Vercel Docs ↗
“Default Maximum Extended maximum Hobby 300s (5 minutes) 300s (5 minutes) - Pro 300s (5 minutes) 800s 1800s (30 minutes) Beta Enterprise 300s (5 minutes) 800s 1800s (30 minutes) Beta”
Read 2026-09-13 · official docs
Docker deployment
Not documented
Node apps routinely need a system binary the platform's buildpack does not ship. A Dockerfile is the escape hatch when they do.
Scales to zero
Supported
An always-on process bills while idle. Scaling to zero removes that cost and adds a cold start to the next request instead.
Sources (1) →Sources ↓
- What is Compute? - Vercel Docs ↗
“You never pay for idle CPU, and you pay nothing between requests.”
Read 2026-09-05 · official docs
What Vercel documents.
Vercel documents a generic path that covers Node.js, rather than a dedicated guide.
Use the Node.js runtime to deploy Node.js HTTP servers or Vercel Functions written in JavaScript or TypeScript. Limited: documented only via the generic Node.js function runtime reference — Vercel publishes no dedicated Node.js app deployment guide (only per-framework guides such as Express).
Sources (1) →Sources ↓
- Using the Node.js Runtime with Vercel Functions ↗
“Use the Node.js runtime to deploy Node.js HTTP servers or Vercel Functions written in JavaScript or TypeScript. Limited: documented only via the generic Node.js function runtime reference — Vercel publishes no dedicated Node.js app deployment guide (only per-framework guides such as Express).”
Read 2026-09-05 · official docs
Node.js features on Vercel
Capabilities Vercel documents specifically for Node.js, with the constraint it documents alongside each.
websockets
Limited
WebSockets are available in Beta on all plans [...] WebSockets in Vercel Functions work exactly like any distributed WebSocket server, so you can use libraries like ws to upgrade connections with no additional configuration [...] WebSocket connections close when a Vercel Function reaches its maximum duration
configurable_request_timeout
Supported
The maximum duration configuration determines the longest time that a function can run. You can configure the maximum duration for Vercel Functions. [...] Pro and Enterprise teams can set individual Vercel Functions using supported Node.js, Bun, and Python runtime versions to run for up to 30 minutes.
background_workers
Limited
Send messages for agent events and background work to durable topics, then process them with independent consumers, automatic retries, and at-least-once delivery. [...] Push and poll modes: Process on Vercel with push callbacks or run your own workers.
Sources (3) →Sources ↓
- WebSockets ↗
“WebSockets are available in Beta on all plans [...] WebSockets in Vercel Functions work exactly like any distributed WebSocket server, so you can use libraries like ws to upgrade connections with no additional configuration [...] WebSocket connections close when a Vercel Function reaches its maximum duration”
Read 2026-09-13 · official docs
- Configuring Maximum Duration for Vercel Functions ↗
“The maximum duration configuration determines the longest time that a function can run. You can configure the maximum duration for Vercel Functions. [...] Pro and Enterprise teams can set individual Vercel Functions using supported Node.js, Bun, and Python runtime versions to run for up to 30 minutes.”
Read 2026-09-13 · official docs
- Vercel Queues ↗
“Send messages for agent events and background work to durable topics, then process them with independent consumers, automatic retries, and at-least-once delivery. [...] Push and poll modes: Process on Vercel with push callbacks or run your own workers.”
Read 2026-09-13 · official docs
What it costs to run there.
- Entry paid plan
- $20 per developer seat / month (Pro)
- Pricing model
- Per-seat subscription plus metered resource usage; CDN charges cover requests, data transfer to visitors and Fast Origin Transfer, and vary by the region a request comes from.
- When you exceed a cap
- Billed automatically: extra usage beyond the included amount or Pro credit is charged, not throttled or suspended
- Regions
- 20 compute-capable regions, from Stockholm and Mumbai to Cleveland and Cape Town; functions default to iad1 (Washington, D.C.).
Sources (4) →Sources ↓
- Vercel Hobby Plan ↗
“Developer seats cost $20 per user / month, while Viewer seats are free”
Read 2026-09-05 · official docs
- CDN — Vercel Docs ↗
“CDN pricing covers CDN requests, data transfer to visitors, and Fast Origin Transfer. Pricing can vary by the region where requests originate. CDN requests appear as Edge Requests in Billing and on the Usage page.”
Read 2026-09-14 · official docs
- Pricing on Vercel ↗
“Some resources include an amount of usage your projects can use within your billing cycle. Pro teams also receive a monthly usage credit and can use many resources on demand. If you exceed an included amount or credit, Vercel charges for the extra usage.”
Read 2026-09-05 · official docs
- Regions — Vercel Docs ↗
“Vercel regions: Behind these PoPs, we maintain 20 compute-capable regions where your code can run close to your data. [...] Vercel Functions default to running in the iad1 (Washington, D.C., USA) region.”
Read 2026-09-14 · official docs
Whether it suits you.
Teams building on a framework the CDN understands natively, who want a predictable per-seat cost with metered usage on top. The Git integration covers GitHub, GitLab, Bitbucket and Azure DevOps, so it drops into an existing review workflow.
Consider something else if…
The team is large enough that $20 per developer seat dominates the bill rather than the metered usage underneath it.
Visit Vercel documentation