Work with us
We build RL and evaluation environments that emulate real-world complexity and focus on robust, safe grading for labs looking to train the best agents.
Pre-Deployment Model Evaluations
Safety and capability evaluations through anonymized checkpoints on Multi-Agent Arena, and our internal environments that emulate real-world alignment scenarios.
Real-World Agentic Capabilities
Complex simulations of real-world settings like enterprises with many services, game development, computer use, and more, with care for reward hacking and eval awareness in grading.
Alignment Environments
Simulated scenarios across multi-agent social games and real-world enterprises that test and train models for alignment.
Get in touch
Talk to our team about what you are looking for. We regularly work with research teams, so we are deeply familiar with the samples, standards, and communication that people working with us need to do work at the frontier.