E2B

Arena started in 2024 as a UC Berkeley Sky Computing Lab project for comparing models head to head. Today, it's grown into a platform for evaluating model performance across agents, text, code, images, and videos, running up to 600,000 E2B sandboxes a day. Each sandbox is an isolated cloud computer where an agent can write code, install dependencies, and work for hours on coding, research, reports, and presentations. That isolation keeps results trustworthy and secure: no session's code or files can reach another's and skew the comparison. That security at scale gets tested every time a frontier model drops. Before GPT-5 went public, people rushed to Code Arena to try it first. "When GPT-5 was about to come out, a lot of people came to Code Arena to experience the model firsthand because it wasn't out to the public yet. E2B was the backbone behind all of that. So when we had this massive surge, we didn't have to worry about whether we could handle it." - Aryan Vichare, Founding Engineer, Arena Read the full case study linked in the comments.

Introducing E2B Embed: it brings AI agent sandboxes directly into your product and run them on infrastructure you control. E2B Embed packages the full E2B stack on a single node, giving you the flexibility to deploy in your own environment or a customer’s. Open source under Apache 2.0, it supports Docker Compose, Terraform on GCP, and Kubernetes. All with the same E2B SDK, CLI, and API. https://lnkd.in/gBvCAyaD

Behind every great agent is a great stack. We’re bringing E2B, Fireworks AI, 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 (registration link in the comments)↓

Why make an agent choose one way of solving a problem when it can explore them all? Catch our very own Mish Ushakov and Ondrej Drapalik at AI Engineer Paris 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.

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

A shared heap is a shared risk. This is why we built E2B with customized Firecracker microVMs, each with its own kernel. Give your agent’s untrusted code a separate sandbox with E2B.

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 ↓ https://lnkd.in/g2Ni9s-w

🚀 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. Link in the first comment.

Go from “we should build that” to a working agent faster with the Agents API. Now you can build on the same harness that powers Codex. We handle orchestration, long-running sessions, and context management, so your team can spend more time building something customers want. You define your agent’s tools and workflows, and choose where it runs code and works with files. Start with an OpenAI-hosted sandbox, bring your own environment, or connect a sandbox provider. Plus, we keep improving the harness as new model capabilities arrive. Less infrastructure upkeep on a roadmap that’s already full. Available in public beta: https://lnkd.in/e79vVXGM