E2B for RESEARCH
One-time $20,000 in E2B credits and the Pro tier for academic and nonprofit research teams. Run evals, RL rollouts, and benchmarks in isolated machines, in parallel.
The Pro tier,
and $20,000 to spend on it.
$20,000 in credits
One-time. Spend it on evals, RL rollouts, benchmarks, or code-generation research.
A full machine for your agent
Shell, filesystem, network, and an optional desktop. One isolated microVM per session.
Works with any agent stack
Any model, any framework. The runtime is Apache-2.0, so there is no lock-in.
Custom sandbox CPU and RAM
Pick the CPU and RAM per sandbox. The default is 2 vCPU and 4 GiB.
100 concurrent sandboxes
For parallel agents, evals, and test runs. Expandable to 1,100 with an add-on.
Sessions up to 24 hours
Long tasks stay alive. Pause a sandbox for free and resume it from the same state.
Research teams
in the program
Program
requirements
Affiliated with a university, public research institute, or research nonprofit
First-time E2B users
Every experiment gets a machine
Common questions,
answered.
Early-stage startups, and research teams at universities, public research institutes, or nonprofit research organizations. Both programs are for teams building or studying AI agents: coding agents, data analysis, evals and benchmarks, computer use, deep research, general agents. The eligibility criteria are listed above.
Apply
today
Meet the criteria above? Send the form. We reply once it is reviewed.