E2B

E2B is an open-source platform that runs AI-generated code in secure, isolated cloud sandboxes powered by Firecracker microVMs, giving AI agents a virtual computer to execute code and use real-world tools.

E2B (legal entity FoundryLabs, Inc.) provides secure, isolated cloud sandboxes designed for AI agents and AI apps. Each sandbox is a Firecracker microVM that can run untrusted, LLM-generated code, execute terminal commands, install packages, manipulate files, and control a virtual desktop. The platform is LLM-agnostic and works with any model (OpenAI, Anthropic, Mistral, Llama, and others). E2B ships SDKs for Python and JavaScript/TypeScript, along with a Code Interpreter SDK and a Desktop Sandbox for computer-use agents. Sandboxes start quickly, can run for up to 24 hours, and support custom templates. E2B is open source and offered as managed cloud, bring-your-own-cloud (BYOC), and self-hosted/on-premise deployments for enterprises with data-residency and compliance needs.

Category

AI Agents & Orchestration

Pricing

Usage-based, Free tier, Subscription, Enterprise

Target Market

Enterprise, Startups, Developers

Founded

2023

Headquarters

San Francisco, California, United States

• ✓
  Firecracker microVM sandboxes

  Secure, isolated cloud sandboxes powered by Firecracker microVMs, built to run untrusted, LLM-generated code with fast startup and sessions up to 24 hours.

• ✓
  Code Interpreter SDK

  Python and JavaScript/TypeScript SDK for safely executing AI-generated code, running terminal commands, installing packages, and handling files inside a sandbox.

• ✓
  Desktop Sandbox

  A secure virtual desktop (GUI computer) in the cloud for LLMs, enabling computer-use agents to click, type, and browse.

• ✓
  LLM-agnostic and framework integrations

  Works with any model provider and integrates with agent frameworks such as LangChain, LlamaIndex, and AutoGen.

• ✓
  Custom sandbox templates

  Fully customizable environments via custom templates or runtime package installation, so any Linux-runnable workload can run in the sandbox.

• ✓
  Flexible deployment (cloud, BYOC, self-hosted)

  Deploy on E2B's managed cloud, within your own cloud (BYOC), or self-hosted/on-premises, with US and EU regional options for data residency.

• •
  AI data analysis and code interpreter

  Give AI apps a sandbox to run generated Python/JS for data analysis, charting, and computation without exposing host infrastructure.

• •
  Autonomous coding agents

  Provide coding agents an isolated environment to write, execute, and test code, install dependencies, and inspect errors.

• •
  Computer-use agents

  Use the Desktop Sandbox to let agents operate a virtual computer, browse the web, and interact with GUI applications.

• •
  Reinforcement learning and evaluation

  Spin up many concurrent sandboxes to run and evaluate agent trajectories at scale.

On top of plan fees, E2B bills usage-based per second of a running sandbox based on compute resources (roughly $0.000014/s per vCPU and $0.0000045/GiB/s of RAM, per the pricing page), with some free storage (10 GiB Hobby, 20 GiB Pro). Enterprise pricing is custom. Verify current rates on e2b.dev/pricing.