Case Study: Laude Institute achieves 100x experiment throughput with Daytona

Laude Institute runs 37k sandboxes in a week with Daytona

Laude Institute, a leading AI research institute co-founded by Andy Konwinski, faced a significant challenge in scaling its AI agent benchmarking framework, Terminal Bench. Their initial infrastructure relied on local Docker containers, which severely limited them to running only four to six experiments simultaneously. This bottleneck prevented them from meeting the demands of research labs needing to run thousands of parallel experiments. To overcome this, they turned to the vendor Daytona for its managed runtime platform.

By implementing Daytona, Laude Institute was able to provision thousands of isolated, parallel sandboxes on-demand in the cloud. The solution provided immense scalability, enabling the creation of 37,000 sandboxes in a single week. This resulted in a 100x increase in experiment throughput compared to their previous Docker setup and led to a 30% faster task completion rate. Daytona's platform and dedicated support completely transformed Laude's operations, allowing them to confidently scale their infrastructure to support global AI research teams.

Laude Institute