Nat Friedman, former CEO of GitHub, created a bounty for a playground to test and compare large language models (LLMs). He posted it on Replit Bounties, a marketplace where developers can work with creators from the Replit community. The project was funded by co-founders Matt Huang and Alexandr Wang, and was won by developer Zain Huda, who created an initial version of the playground that included features like text streaming, highlighting, and temperature controls. After implementing core components, another developer Alex Lourenco joined to implement model comparison, hosting, and final polishes. The project became popular, with the website becoming the most popular LLM-comparison tool in February 2023. The code was later open-sourced by Nat Friedman to allow others to build their own playgrounds.