# What is serverless computing?

Serverless computing runs your code without servers to manage: the platform starts it on demand and bills only while it works.

**A Runtime sandbox bills the way serverless does, CPU only while it is busy,
but it is a whole Linux machine that keeps its memory, files and processes
between calls.** Waiting on a model, a 2 vCPU, 4 GiB sandbox costs
$0.03125 an hour; after 60 seconds with nothing happening it pauses itself
and the next request wakes it ([pause when idle](/glossary/idle-pause)).

## Where the term comes from

The CNCF Serverless Working Group's whitepaper defines serverless computing as
"the concept of building and running applications that do not require server
management". Code is packaged as functions that are "executed, scaled, and
billed in response to the exact demand needed at the moment", and there is "no
charge when code is not running". AWS Lambda is the reference example: it bills
duration by the millisecond and a price per million requests.

## What serverless gives up

| Need                               | A function                       | A sandbox on Runtime                                        |
| ---------------------------------- | -------------------------------- | ----------------------------------------------------------- |
| Run untrusted, generated code      | Usually refused, or one language | Any language, `sudo`, its own kernel                        |
| Keep state between calls           | Reload it each time              | Files, memory and processes stay                            |
| Long jobs                          | Capped per call                  | As long as the lease, or `persistent` while credit lasts    |
| A shell, a desktop, a port to open | No                               | Yes                                                         |
| Pay nothing while idle             | Yes                              | Paused storage only, $0.08 per GB a month |

That gap is why agent platforms run tools in sandboxes rather than functions:
an agent's work is stateful, runs code nobody reviewed, and waits long
stretches between steps.

## How Runtime gets close to scale-to-zero

- **CPU is measured.** An idle sandbox pays its memory and a floor of a twentieth of a vCPU, not every vCPU it holds ([CPU floor](/glossary/cpu-floor)).
- **Idle pause.** After 60 seconds with no request, command, connection,
  traffic or CPU use, it pauses and stops paying for compute.
- **Wake on request.** The next command, file call or preview visit wakes it;
  a paused sandbox ran its next command 153 ms after that call at the median on Runtime's servers
  on 28 September 2026 ([speed](/docs/speed)).

## Related

- [What is a cold start?](/glossary/cold-start)
- [What is idle pause?](/glossary/idle-pause)
- [What is a warm pool?](/glossary/warm-pool)
- [What is per-second billing?](/glossary/per-second-billing)

## Sources

Checked 27 September 2026.

- [CNCF WG-Serverless whitepaper](https://github.com/cncf/wg-serverless/blob/master/whitepapers/serverless-overview/README.md)
- [AWS Lambda pricing](https://aws.amazon.com/lambda/pricing/)