An open-source large language model developed in Beijing just shot to the top of the leaderboard for front-end coding tasks.