One billion sandboxes created on E2B. Every one of them powered an AI agent solving a real-world problem. From early-stage startups to cutting-edge AI labs to the Fortune 100, E2B is the infrastructure behind some of the most widely used agents in production, and the RL environments used to train frontier models. Come build with us.

I used E2B Runtime as the foundation for Brezel instead of rebuilding the Firecracker stack. I focused on durable workspaces, lifecycle recovery, and a secure node-local path, then patched CPU placement with cgroup isolation, physical-core pinning, and guest SMT disabled. That cut observed vCPU migrations from ~1,854–2,292 per thread to [0, 1, 1, 1]. On Starsling’s public Node workload, Brezel reached 29.90 runs/s across five fresh 4-vCPU / 8-GiB sandboxes, 5.9% above the published leader. github.com/infercrane/brezel

i put jev @typesafeai in charge of air traffic in the bay area 900 live flights, watched from inside an @e2b sandbox. every risky pair judged in 103ms, zero conflicts missed. gpt-4o missed 56 in the same sky, and opus 5.5 needed 12 sec to just to get 0 github.com/7datumba/jev-towe…

2,000,000 weekly downloads 🚀

another demo🤯! this time a coding agent using @e2b and @rebunoio. the tools run in a e2b sandbox and rebuno handles the policies and durable execution in the demo, i kill the worker mid task, restart it, and the run picks back up in the same sandbox opening the pr waits for approval, and the sandbox pauses until its approved

VMs getting their moment at Meta Connect 👀 Loved hearing @Meta highlight their importance for running Muse agents securely. It’s why every E2B sandbox gets its own microVM and kernel. 🔒

besides carbon, here's some of my favorite software that offers BYOC 1. @planepowers (jira/linear) 2. @turbopuffer (search) 3. @e2b (sandboxes) 4. @ClickHouseDB (olap) 5. zero by @rocicorp (sync engine)

tfw you know something about your company before investors do. Feels like having a special attack fully charged right before you unleash it. 🚀 📈

OpenAI's Agents API gives every agent an environment for files and code … for developers. JoyStream Agent Sandbox does the same for developers and the organizations that have to take those agents to production. A JoyStream agent gets a fresh microVM on @e2b when it needs to run code. Created on the first step that needs it, gone when the run ends, every command and every file it produced in the run ledger. No credential ever enters the VM. Not a connector secret, not a model key. Every outside call goes through the JoyStream MCP gateway instead. Zero trust by construction: there is no setting that lets a secret in. How it fits into a run, what goes in, what comes out, and a side-by-side with the OpenAI Agents API: joystream.ai/blog/joystream-…

Behind every great agent is a great stack. We’re bringing @e2b, @thefireworksai, and @braintrust together in SF for a night on: → sandboxes → inference → observability → what it actually takes to run agents in production Lightning talks, 🍣 sushi, drinks, and builders. Come get STACKED ↓ luma.com/e2b-0e34

Make your cloud the most secure place for AI agents to run. E2B security. Your cloud. Your rules.

Jev is crazy ! on how fast it can detect my gestures in milliseconds faster than my brain thinking about what I should do with current element in my computer. I just have a chance to try Jev on Workser desktop that we built for computer use agent before Jev coming. and Jev is changing the game forever. I think Workser will launch computer use agent with a mixed of LLM and Jev in one optimized experience soon. Free to use if you BYOK. You can also talk with it, will demo tmr. Not just use your computer but 4 parallels computers for ai agents to work on: - local foreground (your own computer) - local background (@trycua) - cloud computer (@e2b) - sandbox (@e2b) It will have command center for you to work with AI Computer Workforces in parallel managed by one voice agent/normal text ai agent like Tony Stark with Jarvis in iron man scenes cuz I'm inspired by that haha. Thanks Jev and GPT Live 1 ! Such a great time to be alive !

People seem to be getting excited about sandbox runtimes with Kubernetes-friendly packaging (not scheduling!). I think some very exciting stuff is coming out soon from @e2b

Why make an agent choose one way of solving a problem when it can explore them all? Catch our very own @mishushakov and @ondrej_sh at @aiDotEngineer on Thursday, September 24, as they show how agents can fork live machines to explore multiple paths in parallel- each with its own copy of the machine’s state, filesystem, and running processes.

Behind every production agent is a serious stack. Join @e2b, Fireworks, and @braintrust on 9/30 for technical talks on sandboxing untrusted code, serving models at agent-loop speed, and knowing whether a change actually helped. RSVP: luma.com/e2b-0e34

Testing AI-written code is one thing, testing it against real customer data raises the stakes. How do you give testing agents the access it needs while keeping that data isolated? Lark built an AI test platform that maps a customer's product, writes end-to-end tests, and keeps running them as the code changes. That means testing environments with real user data and credentials in play. Lark also needed to run Docker inside the sandbox itself to spin up a customer's own dev environment for testing. @e2b's MicroVM isolation gave Lark strict security boundaries plus Docker-in-sandbox support, without having to build that isolation layer themselves. "They're testing their live environments with real customer data or real API keys, and so everything has to be securely isolated. We don't want an agent in one sandbox to be peeping into another sandbox. It's crucial to the security of our product." - Jack Brown, CEO, Lark @uselark_ai Read the full case study: e2b.dev/customers/lark-case-…

E2B is now on Highway 101. Every agent gets a machine. Every driver gets the memo.

We’re hiring a community builder to make E2B the best place on the internet for people building agents. Talk with builders on X. Grow our Discord + GitHub. Bring everyone together for workshops, dinners, and hackathons. Come build the agent community with us ↓ jobs.ashbyhq.com/e2b/648efda…

We open-sourced our infra repository in the very early days of @e2b Today we’re reintroducing it as E2B Runtime: the open-source execution layer for agents running on E2B.

🚀 Our open-source project has a new name: E2B Runtime. Behind every E2B stack there's a runtime. Cloud, enterprise, or embedded in your product. Same code, Apache-2.0. Old URL redirects. Something new is already in the tree. More soon. github.com/e2b-dev/runtime

Behind every great agent is a great stack. We’re getting the best one together for a night that’s stacked AF. Join @e2b, @FireworksAI_HQ, and @braintrust for technical talks, drinks, and bites on 9/30. RSVP: luma.com/e2b-0e34