On November 6th, ZeroOneAI, led by Kai-Fu Lee himself, officially open-sourced two large pre-trained models, Yi-34B and 6B, on their website. Yi-34B has achieved significant success on the global open-source large model leaderboard, becoming a dual champion, especially excelling in Chinese metrics. Yi-34B supports a version with an ultra-long context window, capable of processing around 400,000 Chinese characters of super-long text input, which is of great help to fields such as law, finance, and marketing. ZeroOneAI plans to release a series of quantized versions, conversational models, mathematical models, code models, etc., to collectively promote the development of the language model open-source community.