Luchen Technology announced a hundreds of million series A++ funding round. In 2021, ZhenFund led the seed round in Luchen Technology.

Founded in 2021, Luchen Technology embraces the vision of "unleashing AI productivity" and leverages numerous technologies in the field of high-performance computing to accelerate and enhance the research and development, as well as application, of large models. Luchen's large model training and inference system, Colossal-AI, and its video large model, Open-Sora, have both topped the global rankings on Github, leading the world in open-source indicators for AI training and inference systems as well as video models. Collectively, they have garnered 60,000 stars on GitHub. Luchen has also achieved remarkable results in inference system optimization, with its open-source inference system's latency comparable to top technologies like vLLM and JAX, and a 50% increase in inference speed.

Founder and CEO Yang You, who was one of the participants in the ZhenPlanet, stated, "Why do we believe that video large models are more likely to have long-term scaling laws than LLMs? Because the training data for video large models is a true reflection of the objective world. The ultimate creator of this data is the Creator itself, and large models will surely uncover its inherent laws. The training data for LLMs comes from the internet and books, where the level of data creators varies greatly, and there is much ambiguity and junk information.

Most importantly, the field that will demand the highest optimization of computing power in the future is likely to be video large models. The super apps of video large models will ultimately compete on computing power efficiency, which is where our strength lies."

Following this round of funding, Luchen Technology will focus on developing: 1) A new DataBricks platform for the era of large models, maximizing the post-training and inference efficiency of LLMs; and 2) World-class large models for video generation.