E2B
Sandboxed cloud machines that AI agents run code in
Overview
E2B runs AI-generated code in isolated cloud microVM sandboxes that can be paused, snapshotted and forked, with Python and JavaScript SDKs and per-second billing. Used by Hugging Face, Perplexity and Manus; free Hobby tier with a one-time $100 credit, Pro from $150 a month, and enterprise bring-your-own-cloud.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 1 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- $100 starter credits
- Up to 1-hour sandbox sessions
- Up to 20 concurrent sandboxes
- Customizable sandbox CPU & RAM
- Up to 24-hour sandbox sessions
- 100 concurrent sandboxes (600 for +$500/mo, 1,100 for +$1,000/mo)
- Multi-day and persistent sessions
- Tens of thousands of concurrent sandboxes
- Bring-your-own-cloud deployment
Use cases
What you can produce with E2B
- Isolated cloud microVM sandboxes for AI-generated code execution
- Python and JavaScript/TypeScript SDKs, plus Code Interpreter and Desktop sandboxes
- Pause, resume, snapshot and fork sandboxes
- Per-sandbox network firewall and secrets vault
- Apache 2.0 runtime; single-node Embed for evaluation, BYOC for production
- Per-second usage billing on top of subscription tiers
E2B is plumbing, and good plumbing: isolated sandboxes purpose-built for letting an AI agent run arbitrary code, browse or use a desktop without putting your own systems at risk, with SDKs and documentation that make it quick to wire in. Snapshots, forks, a network firewall and a secrets vault have made it more capable through 2026, and its customer list — Hugging Face, Perplexity, Manus, Groq — plus built-in support in OpenAI's Agents SDK shows it is widely used. It is not something end users touch; the value appears once you build a code interpreter, data-analysis agent or coding tool on top. Pricing is $150 a month for Pro plus per-second compute, so costs scale with how hard your agents work, and the free Hobby tier's one-hour sessions are for prototyping. Self-hosting is evaluation-only unless you buy enterprise BYOC. Compare it with Daytona and Modal by building on the SDK, since there are no independent reviews.
What we found
E2B, run by FoundryLabs Inc., sells cloud sandboxes for AI agents: isolated virtual machines in which an agent can run code, use a shell, browse or drive a desktop without touching your own systems. Its SDKs cover Python and JavaScript, with a code-interpreter library that runs model-written code and a desktop sandbox for computer-use agents. Sandboxes can be paused, resumed, snapshotted and forked, up to a hundred forks per request, and newer controls add a per-sandbox network firewall and a secrets vault whose values never enter the sandbox. It now calls itself "the enterprise AI agent cloud", lists SOC 2 Type II and HIPAA compliance, and runs in US, EU and APAC regions; it does not say which APAC country.
Adoption claims are the company's own but large: 94 of the Fortune 100 signed up, more than 100,000 teams and over a billion sandboxes started. Named customers include Hugging Face, Perplexity, Manus, Groq, Lindy, Genspark and LM Arena; Manus says its agent uses 27 different tools inside E2B. It is also built into OpenAI's Agents SDK. E2B has raised $32 million, including a $21 million Series A led by Insight Partners in July 2025; founders Vasek Mlejnsky and Tomas Valenta are its chief executive and chief technology officer. A January 2026 control-plane outage was followed by a published postmortem.
The core is Apache 2.0 on GitHub, and the repositories are actively maintained. Self-hosting is narrower than often described: the new E2B Embed package runs everything on one machine and is explicitly for evaluation, while production inside your own AWS, GCP or Azure account is a bring-your-own-cloud deployment that E2B operates for enterprise customers.
Pricing is a plan plus per-second usage. Hobby is free with a one-time $100 credit, one-hour sessions and 20 concurrent sandboxes. Pro costs $150 a month for 24-hour sessions, custom CPU and memory and 100 concurrent sandboxes, expandable to 1,100 for up to $1,000 more. Enterprise has a $3,000 monthly minimum. Comparisons online mostly come from rival sandbox vendors, and there are no G2 or Capterra reviews.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including E2B's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with E2B unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to E2B directly →
Spotted something out of date? Suggest an update →
Alternatives to E2B
More in Code & Dev Tools
