Every agent needs a computer
Sailboxes are persistent Linux VMs specifically designed to be
the cloud environment for long-horizon AI agents.
Equipped for the
long horizon
The first purpose-built sandbox for AI agents that work harder and run longer.
Run your cloud agents in Sailboxes and enjoy the efficiency (and the economics).
Full machines
A full, kernel-isolated Linux VM with root access. Independent disks, persistent state, Docker support, and local NVMe.
Uncapped lifetimes
No runtime limits. Agents can live for days, weeks, or forever without losing state.
Delightful CLI and SDKs
Run Sailboxes with ease using the ergonomic CLI / TUI or SDKs in Python, TypeScript and Rust.
Effortless durability
Pause, resume, and fork all sandbox state; in-flight work survives. Open connections survive for up to 10 minutes.
Over 70% less than other providers.
Long-horizon agents can live for days, weeks, or longer — yet spend most of their time dormant. Sailboxes only charge you for the exact portion of CPU, memory, and disk you consume every second, so these workflows stay scalable and sustainable.
| Sail | $0.015 active vCPU-hr | $0.008 active GB-hr | No fixed limit | <3s | |||
| Modal | $0.071 reserved vCPU-hr | $0.024 reserved GB-hr | 24 hours | Alpha | – | Alpha | <500ms |
| E2B | $0.0504 reserved vCPU-hr | $0.0162 reserved GB-hr | 24 hours | – | <1s | ||
| Vercel | $0.128 active vCPU-hr | $0.0212 reserved GB-hr | 5 hours | – | – | <1s | |
| Daytona | $0.0504 reserved vCPU-hr | $0.0162 reserved GB-hr | No fixed limit | Exp. | – | <500ms | |
| Cloudflare | $0.072 active vCPU-hr | $0.009 reserved GB-hr | No fixed limit | – | – | – | <2s |
Developers ♥ Sailboxes for...
Long-horizon agent products
Quadrillion Labs uses Sailboxes to host cloud agents for their autoresearch platform, Qualia Cloud. Ask a Qualia agent to test a hypothesis and it spins up a fresh Python kernel inside a Sailbox in seconds, running multiple experiments in parallel — coordinating results between agents through our NFS-mounted Volumes.
RL rollouts
Sailboxes are perfectly suited for rollouts leveraging Sail's Tinker-compatible API, with support for any Prime Intellect environment. Sailboxes automatically sleep while an agent waits on a Sail inference call, so achieving max cost efficiency is effortless. Since there's no human in the loop, combining Sailboxes and Sail inference lets you run far more rollouts per dollar.
Devboxes
Because every Sailbox is a full computer in the cloud, it's an ideal place to do development work. Use Sailboxes to host coding agents, run background automations, or even power always-on assistants like OpenClaw and Hermes Agent.
Sailboxes + Sail inference
Sleep while you think
Agents spend a lot of time waiting; don't pay for this. Sailboxes automatically sleep while waiting on Sail inference calls.
One efficient stack
Use Sailboxes and Sail inference together to achieve the most cost-efficient and scalable AI agents on the leading open source models.
Two ways to Sail.
Spin up in seconds and let the tokens flow.
Get started with $5 free creditWorkloads at extreme scale.
Talk to usPutting the Sailbox's live migration mechanism to the test by running a Minecraft server in a Sailbox.
Learn how Sailboxes Work