Back to vendors
Entry priceFree Hobby ($100 one time credit) · Pro $150/mo + per second usageFull pricing detail

Secure code sandbox infrastructure for AI agents: isolated virtual-computer execution environments, SDKs, and PAYG pricing for AI developers who need safe, ephemeral code execution. Infrastructure component, not an end-user agent.

E2B is open-source cloud infrastructure that gives AI agents secure computers to work in. Its core idea addresses a basic problem with agentic systems: an agent that writes and runs code needs somewhere to execute it, but running untrusted, AI-generated code on your own machines or servers is dangerous. E2B provides isolated, disposable cloud sandboxes where an agent can run code, browse the web, and operate a desktop without ever touching your own infrastructure.

Each sandbox is a fast Linux virtual machine that spins up on demand in milliseconds and can persist its filesystem and state across a multi-step task, so an agent can install packages, write and edit files, run commands, and pick up where it left off. E2B builds these on Firecracker microVMs, the same lightweight virtualization used for serverless computing, which gives a stronger isolation boundary than ordinary containers, an important property when the code being executed was generated by a model and cannot be trusted.

The platform is deliberately model- and framework-agnostic, working with any major LLM and any language. Developers drive it through Python and TypeScript SDKs and a CLI, with a code-interpreter SDK for running model-generated code and capturing its output, and an E2B Desktop variant that gives an agent a full graphical Linux desktop to see and control for computer-use tasks. Because everything is open source, including the underlying infrastructure, teams can also self-host.

A defining strength is concurrency and scale: E2B is built to launch large numbers of sandboxes in parallel, which makes it a common substrate not just for coding agents but for running many agent instances at once, evaluating reinforcement-learning reward functions across many environments, and powering data analysis and research workflows. It is used as the execution layer behind well-known AI products and autonomous agents, and offered as a hosted service as well as bring-your-own-cloud and on-premises deployments for enterprises with security or requirements. In short, E2B is the secure runtime that lets agents act on the world through code, without putting the host environment at risk.

Vendor details

Canonical URL

https://e2b.dev

Category

Agent infrastructure

Company status

independent

Use cases & customers

Target customers

AI developersinfrastructure teams

Deployment options

SaaSAPI

In practice

Your agent writes Python and you can't safely run it on your own servers. E2B spins up an isolated Firecracker microVM in milliseconds where the model-generated code executes, then returns the results without touching your infrastructure.

You're running dozens of coding agents in parallel to speed up your team. E2B launches many isolated sandboxes at once, each a full Linux environment, so the instances run side by side without colliding.

Your agent needs to click through a real desktop app, not just call APIs. E2B Desktop gives it a full graphical Linux desktop to see and control, providing a safe sandbox for computer-use tasks.

Sources & related URLs

Agentic Index coverage score

7.0 / 14 capabilities · 50%

Integrations & Tool CallingSDK and API docs 2026-06-08 Full
Workflow OrchestrationSandbox infrastructure layer Unable to verify
Knowledge Grounding & RAGSandbox infrastructure layer Unable to verify
Human Oversight & GuardrailsSandbox infrastructure layer Unable to verify
Security, Identity & GovernanceIsolated sandbox security docs 2026-06-08 Full
Observability & AuditabilityOfficial docs 2026-06-08 Partial
Memory & State PersistenceEphemeral sandbox model Unable to verify
Deployment & Data ResidencySaaS API docs 2026-06-08 Full
Prebuilt Agents, Templates & PacksSandbox infrastructure layer Unable to verify
Triggers & Channel CoverageSandbox infrastructure layer Unable to verify
Model Flexibility & RoutingModel-agnostic execution docs 2026-06-08 Full
APIs, SDKs & MCP ExtensibilitySDK and API docs 2026-06-08 Full
Testing, Debugging & OptimizationTesting and debugging execution docs 2026-06-08 Full
Browser & Computer UseVirtual-computer framing docs 2026-06-08 Partial

The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Recent platform changes

2026-07-01·MCP / tool calling / APIVerified

E2B deprecated E2B_ACCESS_TOKEN authentication in favor of E2B_API_KEY. This is a breaking change for existing integrations: legacy access tokens stop working on August 1, 2026, and all SDK and API calls must migrate to API-key auth before then.

Bears on: MCP / tool calling / API

View source
2026-06-10·Funding / partnershipVerified

E2B joined the Stripe Projects developer preview so agents can discover, provision, and authenticate E2B sandboxes without manual API-key setup.

Bears on: Integrations

View source
View all 2 changes for E2B →Tracked since Jun 2026 · Verified from public vendor sources

Pricing

Free Hobby ($100 one time credit) · Pro $150/mo + per second usage

usage

Free tier

Included quota

Hobby: one time $100 usage credit, 1 hour max session length, 20 concurrent sandboxes, 10 GiB storage. Pro: 24 hour sessions, 100 concurrent sandboxes (expandable to 1,100 on request), customizable 1 to 8 vCPU and up to 8 GiB RAM, 20 GiB storage. Compute rates: $0.000028/s for 2 vCPU (~$0.10/hr) and $0.0000045/GiB/s RAM, identical on both plans.

What is public

E2B publishes full pricing: free Hobby tier with a one time $100 credit, Pro at $150/mo unlocking production limits, per second compute rates with a public calculator, and custom Enterprise with BYOC, on prem, and self hosted options. The core runtime is open source (Apache 2.0).

Billing mechanics

Per second billing for running sandboxes (CPU and RAM metered separately) layered on a flat plan fee. Billing stops immediately on pause, kill, or timeout. Postpaid monthly.

Cost watchouts

The $150/mo Pro fee is a fixed floor paid before any compute: it buys limits (session length, concurrency), not credits. Spiky or low volume workloads pay the full floor regardless. No GPU option at any tier. Paused sandboxes stop compute billing but storage persists.

Variable cost rationale

Per second metering means cost scales directly with sandbox count and lifetime; long lived or highly concurrent agent fleets can dwarf the subscription fee.

Additional watchouts

Budget the Pro floor against actual utilization: 1,760 sandbox hours a month is good value, 50 hours is not. Teams needing GPU inside the sandbox must run a second platform.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free Hobby tier, no card ($100 one time credit, 1 hour max sessions)

Lowest paid plan

Pro $150/mo plus per second compute usage

Commercial notes

Raised $34.5M+ including a $21M Series A led by Insight Partners (Jul 2025). Customers include Perplexity and Hugging Face; claims 88% of Fortune 100 signed up. Firecracker microVM isolation is the security differentiator. Startup program bundles Pro plus $20K credits.

Key ambiguities

Enterprise pricing unpublished. Concurrency beyond 100 is on request pricing.

Missing data

Enterprise pricing and above 100 concurrency rates unpublished.

Agentic Index verified 2026-07-06

Alternatives to E2B

The closest documented capability profiles to E2B among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • AgentOps6.0 / 14Fuller documented coverage on Observability & Auditability
  • Beam8.0 / 14Adds documented Workflow Orchestration and Memory & State Persistence, among others
  • LiteLLM7.0 / 14Adds documented Human Oversight & Guardrails
  • Modal8.0 / 14Adds documented Workflow Orchestration and Memory & State Persistence, among othersE2B vs Modal →
  • OpenRouter5.0 / 14A lighter documented profile than E2B
  • Opik7.0 / 14Adds documented Human Oversight & Guardrails

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.