Daytona just hit 68K stars for running AI code safely

Everyone is building apps with AI now. Claude Code writes your backend. ChatGPT drafts your frontend. Cursor auto-completes everything in between. But here's the part nobody talks about: who checks if all that AI-generated code is actually safe to run?

Daytona just crossed 68,000 GitHub stars, making it one of the most popular open-source projects for running AI-generated code in isolated environments — called sandboxes (safe, throwaway computers that disappear when you're done) — that spin up in under 90 milliseconds.

Why vibe coders need a safety net

When you ask AI to write code, you're trusting a model that doesn't fully understand your system. The code might work perfectly. Or it might accidentally delete files, open security holes, or crash your machine. You won't know until you run it.

Daytona solves this by creating a disposable computer for every piece of AI-generated code. Think of it like a test room: the code runs inside, and if something goes wrong, nothing on your real machine gets touched. When you're done, the test room disappears.

90 milliseconds from zero to running

The standout number is speed. Daytona creates a fully isolated environment — with its own file system, network, and execution engine — in under 90 milliseconds. That's faster than your screen refreshes.

For comparison, spinning up a traditional virtual machine (a full simulated computer) takes 30–60 seconds. Docker containers (a lighter alternative) take 1–5 seconds. Daytona's approach sits in a different category entirely.

$24M and some heavy backers

Daytona raised a $24 million Series A in February 2026, led by FirstMark Capital. Strategic investors include Datadog (a major monitoring platform) and Figma Ventures (the design tool's investment arm) — two companies that know infrastructure and developer tools inside out.

CEO Ivan Burazin says the company hit $1 million in recurring revenue within three months of launch, then doubled it in six weeks. That kind of growth caught the attention of LangChain, Turing, Writer, and SambaNova, all of which now use Daytona's sandboxes for real projects.

As FirstMark's Matt Turck put it: "The next infrastructure shift is from human-centric cloud primitives to agent-native ones."

Plugs straight into Claude Code, Codex, and LangChain

Daytona isn't just for solo developers. It integrates directly with the AI coding tools most people already use:

This means you can keep using your favorite AI coding tool and add Daytona as a safety layer — without changing how you work.

Try it yourself

Getting started takes three steps:

# Install the Python SDK
pip install daytona

# Or use TypeScript
npm install @daytonaio/sdk

Create a free account at app.daytona.io, generate an API key, and you're ready to sandbox your first AI-generated script:

from daytona import Daytona

daytona = Daytona()
sandbox = daytona.create()

# Run AI-generated code safely
response = sandbox.process.code_run('print("Hello from sandbox!")')
print(response.result)

What's coming next

The team is working on sandbox forking — the ability to clone a running environment's entire state (memory, files, processes) instantly. That means you could test multiple versions of AI-generated code in parallel, each in its own isolated copy.

With 168 releases since launch and a new version dropping every few days (v0.154.0 landed on March 20), Daytona is moving fast. As AI writes more of the world's code, the demand for safe places to test it will only grow.

The project is open-source under the AGPL-3 license and available on GitHub.

Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News