Daytona iconDaytona

Fast, stateful cloud sandboxes for running AI-generated code, with SDKs, Git and LSP APIs, computer-use desktops and per-second pricing

Sub-90ms Sandbox Creation (vendor claim)
$200 Free Compute Credit
2023 Founded

Overview

Daytona runs secure, elastic sandboxes where AI agents execute the code they write. It claims sandbox creation in under 90 milliseconds, and sandboxes are stateful: they can keep running, be snapshotted and get persistent volumes. SDKs for Python, TypeScript, Ruby, Go and Java give agents process, file system and Git APIs, plus LSP support, preview URLs, SSH, a web terminal and a browser VS Code editor. Computer-use sandboxes offer Linux, macOS and Windows desktops, and preemptible GPUs are available. Regions cover the US, EU and India, and Enterprise can run sandboxes on customer-managed compute in your cloud. Daytona started as an open-source dev-environment tool; in June 2026 it announced it is moving its production code to closed source for security reasons, and the old AGPL-3.0 repository is now archived. The Croatian-founded company raised a $24M Series A led by FirstMark in February 2026.

The Verdict

Who Should Use Daytona?

Best For

  • Agent builders who need sandboxes that start fast and stay alive
  • Coding agents that need Git, LSP and a terminal inside the sandbox
  • Computer-use agents on Linux, macOS or Windows desktops
  • Teams that want GPUs and CPUs from one sandbox provider

Not Ideal For

  • Teams that require an open-source, self-hosted sandbox (the repo is archived)
  • Buyers who want a large base of independent reviews before choosing
  • Workloads that need microVM isolation guarantees without checking Daytona's docs first

What's Great

  • Fast sandbox startup is the most repeated praise in comparisons
  • Broad toolkit in one product: Git, LSP, PTY, desktops, snapshots, volumes
  • Stateful, long-running sandboxes with no runtime cap noted
  • Transparent per-second pricing and $200 of free compute
  • Customers include Writer, Clay, n8n, Mintlify and SambaNova

Watch Out For

  • Moving to closed source: the open repository is frozen and archived
  • Self-hosting is not documented after the change; Enterprise BYOC is the stated path
  • One comparison says container-based sandboxes share the host kernel, weaker than microVMs
  • Very few independent reviews on G2 or community forums

Pricing

Billing is per second. Windows sandboxes cost $0.0858 per vCPU-hour. GPUs are preemptible; see the pricing page.

View all features & details

Key Features

  • SDKs for Python, TypeScript, Ruby, Go and Java
  • Process, file system and Git APIs; LSP support
  • Snapshots, volumes, preview URLs, SSH, web terminal
  • Docker and Docker Compose inside sandboxes
  • Computer-use desktops: Linux, macOS, Windows
  • Preemptible GPUs

Use Cases

  • Running AI-generated code safely
  • Long-lived workspaces for coding agents
  • Desktop automation agents
  • Parallel agent runs at scale

Company

  • Founded 2023 by Ivan Burazin, Vedran Jukić and Goran Draganić
  • $24M Series A led by FirstMark (February 2026), with Datadog and Figma Ventures
  • A $60M Series B at a $1B valuation was reported in October 2026 (single source)

How It Compares

Feature Daytona E2B
State Stateful, can run indefinitely Pause and resume, 24-hour Pro cap
Open source No (old repo archived) Yes (Apache 2.0)
Desktops Linux, macOS, Windows Linux
GPUs Preemptible GPUs No
Free credit $200 $100
vCPU price $0.0504 per hour $0.000014 per second (about the same)

Sources: daytona.io · Daytona pricing · Tech.eu (Feb 2026) · Crypto Briefing (Oct 2026) · E2B pricing

Is this your tool? Add a badge