How it works
You write a handler: one function that takes an event and returns an answer. The event can be a web request, a file landing in storage, a message in a queue or a timer. The provider runs your handler when the event arrives, then freezes or removes the environment.
AWS Lambda is a serverless compute service: you run code without provisioning or managing servers. The provider handles server maintenance, capacity provisioning, scaling and patching. There is no upfront commitment, and you pay only for the resources used. On AWS Lambda, 1 million requests and 400,000 GB-seconds a month are free; after that each million requests costs $0.20.
# the whole "server" is this function
def lambda_handler(event, context):
name = event.get("name", "world")
return {"statusCode": 200, "body": "Hello, " + name}
Most of the time nobody calls, so most of the time nothing runs and nothing is billed. That is the appeal. A weekend project or a rarely used webhook can cost close to zero.
What you give up
A cold start is the delay on a call that finds no ready environment: the service downloads your code, starts the environment and runs your initialization code before the handler. AWS says it lasts from under 100 ms to over 1 second, and that cold starts occur in under 1% of invocations. A function that runs rarely meets it more often.
| Serverless function | A computer you keep | |
|---|---|---|
| Longest run | AWS Lambda: 15 minutes per call | No limit while it works |
| Memory between calls | Not kept: built for short tasks that keep no state between calls | Kept: files and running programs |
| First call after a pause | Can be a cold start | Answered at once; if it sleeps, it wakes in a fraction of a second |
| Environment lifetime | Recycled every few hours | Yours until you delete it |
| Who patches the system | The provider | You, or your agent |
| Bill | Requests and run time | A flat plan |
The limits are by design. A function that lives for seconds and forgets everything is easy to copy a thousand times. Serverless is moving: AWS now also sells Lambda MicroVMs, isolated environments for untrusted code that run up to 8 hours per session and keep memory and disk state on suspend. Those are a different product from a plain function.
Do you need it?
Take serverless when
- Calls come in bursts. A webhook that fires a few times a day, then nothing for hours.
- Each call is short and independent. Resize an image, send a receipt, check a form.
- You cannot predict the traffic. The provider adds copies when it grows.
Use a server instead when
- A job runs for hours. An AI agent building a project is cut off at the run limit.
- The program keeps state. A bot with a conversation, a game server, a database.
- You need root. Docker, system packages, your own tools.
The two meet in the middle: a VPS is the server you keep, a function is the server you rent by the second, and this comparison sets them next to a platform service with prices.
Serverless habits on a real computer
An Uno computer is not serverless. It is one Linux machine with root, and the price is flat. It borrows the quiet-time habit: with nothing to do for 10 minutes it sleeps, and memory and running programs are saved while it sleeps. It wakes on any of these: you open it, a visitor opens your site, a message reaches your bot's webhook. Your next request finds everything where it was, and the wake itself takes a fraction of a second.
Small: $5 a month
When a function is not enough, a computer from $5 a month
Every plan is a real Linux computer with root and Docker. It sleeps when idle and wakes on the first request. Move up or down any time.
Small
A Telegram bot, a VPN, a small backend
$5/ month
- 1 vCPU · 1 GB
- 15 GB disk
- 25 GB cloud storage
- 5 boosts a month
Plus
Agents that work with your laptop closed
$20/ month
- 4 vCPU · 4 GB
- 40 GB disk
- 100 GB cloud storage
- Uno Work in the cloud
Pro
A product in production, a second MacBook
$70/ month
- 16 vCPU · 16 GB
- 120 GB disk
- 300 GB cloud storage
- Speed of a MacBook Air M4
Max
A team and many agents at once
$200/ month
- 32 vCPU · 48 GB
- 300 GB disk
- 1,024 GB cloud storage
- 5 public IPv4
Add Uno AI to any plan: 40 h of AI work a month on Small, 100 h on Plus, unlimited on Max. Today: GLM-5.3 Flash and MiMo v2.6 Pro. Or sign in with your own Claude or ChatGPT.