Dedalus Labs

Dedalus Labs provides persistent Linux machines for agent workflows, with fast startup, preserved state, and CLI-based machine control. Pricing is usage-based for active compute, with idle time free and storage included.

AI Инструменты Разработчика AI DevOps Ассистент Агент программирования

Посетить Сайт

Overview

Dedalus Labs provides persistent Linux machines for agent workflows. The product positions these machines as starting in under 250 ms, with filesystem and memory preserved across sessions so work can continue without rebuilding the environment each time.

The site shows a CLI-first workflow for installing the Dedalus client, creating a machine, running commands, and SSHing into the environment. The pricing page adds per-second billing for active compute, free idle time, and storage that persists across sleep and wake cycles.

Core capabilities

Fast machine startup

Machines are described as full Linux environments that start in under 250 ms, reducing waiting time compared with traditional sandboxes.

Persistent runtime state

Filesystem and memory are preserved across sessions, so installed tools, environment variables, and background processes can remain available after sleep and wake cycles.

Flexible Linux environments

The product emphasizes full Linux access with root privileges, support for apt, pip, npm, cargo, and brew, plus GPU/CUDA and Docker support.

Usage-based billing

The pricing page states that compute is billed per second only while the machine is active, with idle time free and storage included in the plan.

CLI-driven workflow

The product shows a command-line workflow for installing the CLI, creating machines, running commands, and connecting through SSH.

Team and enterprise options

The pricing page lists enterprise controls including SSO, RBAC, audit logs, SLA options, and dedicated support.

Practical use cases

Persistent agent runs
Run agents that need a machine to keep state between jobs, including installed packages, files, and background processes.
Flexible development environments
Create a Linux environment for tasks that require system packages, root access, or Docker instead of a limited sandbox.
Pay-for-active-compute workloads
Launch short-lived or intermittent workloads such as command execution, scraping, or code generation without paying for idle time.
Remote machine access
Use a machine as a remote workspace when SSH access and command-line control are part of the workflow.
Team and enterprise fleets
Standardize team infrastructure on a paid tier when centralized controls such as SSO, RBAC, audit logs, or dedicated support are needed.

Pros and Cons

Pros

Starts full Linux machines in under 250 ms, which supports low-latency agent workflows.
Preserves filesystem and memory across sessions, reducing setup repetition.
Supports root access and common package managers, making the environment more flexible than a restricted sandbox.
Charges only for active compute, with idle time billed at $0 and storage included in the plan.
Offers a CLI workflow plus SSH access for hands-on machine use.

Cons

The publicly visible sources do not fully document the API or integration ecosystem.
Team-usage details are only partially described outside the enterprise pricing summary.

FAQ

How do you interact with Dedalus Machines?

The source shows a CLI for creating machines, running commands, and connecting over SSH. It also shows a documentation link and an API key page, but it does not fully document the API surface or integration catalog.

How does billing work?

The pricing page says compute is billed per second only while a machine is actively running. Idle time is free, and storage persists across sleep and wake cycles.

What platforms are supported?

The home page shows support for macOS, Linux, and Windows installation of the Dedalus CLI, with commands for creating machines and SSH access. It does not provide a broader platform compatibility list beyond the CLI instructions shown.

What plans are available?

The pricing page says hobby, pro, and enterprise options are available. Hobby includes 50 compute hours per month and a one-time sign-up credit, Pro includes monthly compute credits, and Enterprise is custom with team controls such as SSO, RBAC, and audit logs.

What kind of workloads is it designed for?

The product is aimed at persistent agent workloads and other workflows that need full Linux machines, preserved state, and root access. The source explicitly highlights machine-based workflows such as command execution and SSH access.