What each one is
| Service | When idle | Longest run | Entry price |
|---|---|---|---|
| Uno | Asleep; memory and running programs are saved while it sleeps | there is no session time limit; it sleeps when idle and wakes with its memory | Small $5 a month, flat |
| E2B | Pause keeps the filesystem and the memory, with running processes and loaded variables, indefinitely, until you kill the sandbox | 1 hour on Hobby, 24 hours on Pro | Hobby $100 credit once; Pro $150 a month plus usage |
| Fly Sprites | Warm: memory frozen. Cold: memory dropped | No session cap stated; pauses after about 30 seconds idle | Pay as you go, no monthly fee |
| Morph | Snapshots; the rules were not readable today | Not checked | Developer $40 a month with 1,000 MCUs |
| Modal Sandboxes | Modal points to filesystem snapshots for state that must outlive a sandbox | 5 minutes by default, up to 24 hours | Starter $0 a month, with $30 of usage credits |
| Daytona | VM sandboxes can pause and resume with memory snapshots; container sandboxes cannot | Not stated | $200 of free compute, then usage |
| Vercel Sandbox | stopping a sandbox saves its filesystem; resuming starts a new session from that snapshot | 45 minutes on Hobby, 24 hours on Pro | Usage on the Pro plan; Hobby has a small free quota |
| CodeSandbox (Together) | snapshots keep memory state; hibernation resume under 1 second, as the vendor says | Runs until you end it | Usage only; access is enabled per organization |
A month of one machine, 1 vCPU · 1 GB class
Whole-month cost in $ per month if the machine is switched on and busy for 730 hours. Shapes differ a little: Morph sells 4 GB with each vCPU, Vercel 2 GB, and Modal bills per core.
Show as a table
| Option | $ per month | Shape |
|---|---|---|
| Uno Small, 1 vCPU · 1 GB | 5 | Flat price, sleeps when idle, keeps memory |
| Vercel Sandbox, 1 vCPU · 2 GB | 31–124 | Memory alone to memory plus a fully busy CPU |
| Morph, 1 vCPU · 4 GB | 37 | Pay-as-you-go MCU rate |
| CodeSandbox, 1 vCPU · 1 GB | 43 | Together AI rates; no plan fee listed |
| Fly Sprites, 1 CPU · 1 GB | 44 | CPU and memory at the published rates |
| Daytona, 1 vCPU · 1 GB | 49 | Compute only, storage left out |
| Modal, 1 core · 1 GiB | 121 | A Modal core is a physical core; runs end within 24 hours |
| E2B Pro, 1 vCPU · 1 GB | 199 | The Pro fee plus compute |
A month of one machine, 4 vCPU · 4 GB class
The same count at four times the size, in $ per month. Morph sells 16 GB with four vCPU and Vercel 8 GB, so those bars are a bigger machine.
Show as a table
| Option | $ per month | Shape |
|---|---|---|
| Uno Plus, 4 vCPU · 4 GB | 20 | Flat price, sleeps when idle, keeps memory |
| Vercel Sandbox, 4 vCPU · 8 GB | 124–498 | Memory alone to memory plus fully busy CPUs |
| Morph, 4 vCPU · 16 GB | 146 | Pay-as-you-go MCU rate |
| CodeSandbox, 4 vCPU · 4 GB | 174 | Together AI rates; no plan fee listed |
| Fly Sprites, 4 CPU · 4 GB | 176 | CPU and memory at the published rates |
| Daytona, 4 vCPU · 4 GB | 194 | Compute only; 4 vCPU is its per-sandbox maximum |
| E2B Pro, 4 vCPU · 4 GB | 344 | The Pro fee plus compute |
| Modal, 4 cores · 4 GiB | 486 | A Modal core is a physical core; runs end within 24 hours |
How we counted
One machine, busy for 730 hours, at the rates below. Storage, network and the free credits that vendors give on signup are left out.
| Service | Published rate | 1 vCPU class |
|---|---|---|
| E2B | $0.0504 per vCPU-hour, $0.0162 per GiB-hour, Pro fee $150 | 730 × $0.0666 + $150 = $198.62 |
| Daytona | $0.0504 per vCPU-hour, $0.0162 per GiB-hour | 730 × $0.0666 = $48.62 |
| Fly Sprites | $0.0385 per CPU-hour, $0.021875 per GB-hour | 730 × $0.060375 = $44.07 |
| CodeSandbox | $0.0446 per vCPU-hour, $0.0149 per GiB-hour | 730 × ($0.0446 + $0.0149), about $43 |
| Modal | $0.141912 per core-hour, $0.024012 per GiB-hour | 730 × ($0.141912 + $0.024012), about $121 |
| Morph | Per MCU: $0.05; one MCU is 1 vCPU-hour with 4 GB of memory-hours, or 5 TB of snapshot-hours | 730 MCUs, about $37 |
| Vercel Sandbox | $0.128 per Active CPU hour, $0.0212 per GB-hour; Active CPU only; time spent waiting on the network or a model does not count | $31 for 2 GB of memory alone, $124 with the CPU busy all month |
Morph: priced at the pay-as-you-go MCU rate; the Developer fee and its 1,000 included MCUs are not applied. Sprites, Vercel and Modal meter what the machine uses, so an agent that mostly waits for a model costs less than these ceilings.
What survives idle
Price is half the question. The other half is what you find when the sandbox comes back: your files, or your running programs too.
For an agent with an open terminal, a loaded dataset or a half-finished build, the first group resumes where it stopped. The last group starts the program again. Both are fine for a one-shot task; only the first fits an agent that works for weeks.
Pay-per-use against a flat plan
Usage billing is cheaper while the machine is busy only part of the month. At E2B and Daytona rates, an hour of 1 vCPU and 1 GB of memory costs $0.0666; at Sprites rates, $0.060375. $5 buys about 75 and 83 busy hours.
Where Uno is worse
- Under about 75 busy hours a month, E2B, Daytona and Sprites cost less than $5. Hobby tiers and signup credits ($200 at Daytona) cover a trial for nothing.
- For many short, throwaway runs started from code, a sandbox service fits better: E2B runs up to 100 at once on Pro, and Daytona says sub 90 ms sandbox creation, as the pricing page says. An Uno computer is ready in about 2 seconds and is meant to stay.
- Morph goes up to 256 vCPUs and 1,024 GB on its Developer plan. Our largest is 32 vCPU and 48 GB.
- Our vCPU are shared, not dedicated, and paid plans run in one region today: Netherlands.
- Wake times are vendors' own figures and our single measurement, not a side-by-side test. We did not read Morph's docs today, so we leave its restore time out.
Your agent gets a clean address of its own
Your computer runs in the Netherlands. Its traffic can leave through a static address of a regular internet provider in one of 8 countries. Only your account works from that address.
- From a data-center address
- Sites see a server range shared with thousands of strangers. The agent's browser hits more captchas and blocks.
- From your Clean exit IP
- Sites see a regular internet provider's address in the country you picked. Fewer captchas, and less reason for Claude or ChatGPT to flag your account: nobody else works from it.
One agent, one computer, from $5 a month
Every plan is a Linux microVM with root and Docker. Pick the size; 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.