E2B

By E2B

Secure, isolated cloud sandboxes that let AI agents run model-generated code safely

Website ↗ Repository ↗ Docs ↗

e2b@2.35.12026-07-22PythonApache-2.0

#sandbox #code-execution #agent #python

13.1k Open source Free tier Paid

Doc version 1.0 · Updated June 28, 2026

Related concepts

Harness Engineering Doc version 1.0 · Updated June 28, 2026 Wrapping an LLM in scaffolding — a control loop, tools, a sandbox, guardrails, and evaluation — so it behaves reliably in production.

Tools Doc version 1.0 · Updated June 29, 2026 The functions that connect a model to the outside world — web search, scraping, browser control, and app integration that bridge to real data and actions beyond the training cutoff.

Sandboxing Doc version 1.0 · Updated June 30, 2026 Running the code an agent writes or pulls in inside an isolated environment — stopping destructive commands, secret leaks, and runaway resource use while still proving the code actually works

Related tools

Browser Use

Browser & Computer Use

OpenAI Agents SDK

Multi-Agent Orchestration

Overview

E2B gives AI agents secure, isolated cloud sandboxes for running model-generated code.
Each sandbox is a fast-booting VM, so an agent can execute arbitrary Python, inspect the output, and keep state across calls without ever touching the host.

flowchart LR
  code[LLM code] --> e2b[E2B Sandbox]
  e2b --> exec[(isolated VM)]
  e2b --> out[stdout / results]

The Code samples tab shows spinning up a sandbox and running code in it.

When to use it

Choose E2B when an agent needs to run code it generated — data analysis, scripts, or tool calls — in a throwaway environment that is isolated from your infrastructure. The open SDK lets you start locally, and hosted sandboxes scale the same code in the cloud.

Pricing

Open SDK, free to start. Hosted sandboxes are billed per second of running CPU, RAM, and storage.

Plan	Price	Notes
Hobby	Free + usage	$100 credits, 20 concurrent sandboxes, 1h sessions
Pro	$150/mo + usage	100 concurrent sandboxes, 24h sessions, custom CPU/RAM
Ultimate	Custom	Tailored rates and enterprise support

Source ↗ · as of 2026-06-27

Example

Python · run code

When to use

Reach for E2B when an agent must execute untrusted, model-generated code without risking the host.

How it flows

run_code ships the snippet to an isolated VM and returns logs, results, and any errors.

Structure

flowchart LR
  code[LLM code] --> e2b[E2B Sandbox]
  e2b --> exec[(isolated VM)]
  e2b --> out[stdout / results]

from e2b_code_interpreter import Sandbox

# Each sandbox is an isolated cloud VM that boots in milliseconds.
with Sandbox() as sbx:
    execution = sbx.run_code("print('hello from a sandbox')")
    print(execution.logs)

    # State persists across calls within the same sandbox.
    sbx.run_code("x = 21")
    result = sbx.run_code("x * 2")
    print(result.text)  # -> 42