Overview
Zidong Taichu is a multimodal model program led by the Institute of Automation of the Chinese Academy of Sciences with the Wuhan Artificial Intelligence Research Institute. Its official site describes a progression from text models to systems that align and reason across image, text, speech, and video inputs. The MindSpore AI repository provides official Taichu model code for multimodal tasks such as image captioning and visual question answering. GitHub lists it under Apache-2.0, but the repository was last pushed in January 2023, so the current Taichu 4.0 service and later releases are not necessarily represented by that codebase.