Run wild. Securely.

AI-native sandboxed runtime for secure, scalable code execution. Built for agents, parallel evals, instant previews and more.

Built for the AI era

Secure your spot to shape the future of AI sandboxing

Run untrusted code safely

Leverage hardware-level isolation powered by Firecracker microVMs for unmatched security.

Optimized for AI workloads

Specifically built to safely and efficiently execute AI agent and LLM-generated code.

Performance on-demand

Effortlessly scale resources to match your application's exact requirements.

Flexible integrations & tools

Quickly integrate and deploy using SDKs across multiple languages and an intuitive command-line interface.

In progress

Keep sessions and state

Seamlessly support long-running tasks and stateful applications with optional persistent storage.

In progress

Granular network control

Easily manage and secure inbound and outbound connections with customizable network policies.

In progress

Simple, powerful, and secure

Easy to use SDKs and CLI, coming soon.

import { Sandbox } from "@tavor/sdk";
const apiKey = "sk-tavor-brHx*******Kpf";
const box = new Sandbox({
apiKey,
timeout: 600,
debug: true,
});
const exec = await box.execute("hostname");
console.log(exec.output);

Frequently asked questions

Can't find the answer you're looking for? Reach out to us at contact@tavor.dev.

Coming soon
Join the waitlist

Secure your spot to shape the future of AI sandboxing

We're putting the finishing touches on our platform that will change how AI-generated code is executed. Join our waiting list today for early access and exclusive benefits.

We respect your privacy. No spam, just updates about our launch and early access.