According to CoinMeta, Tether recently open-sourced a visual language model with 460 million parameters, designed to run directly on smartphones without relying on cloud servers. This model was released by the AI research department's QVAC, and is named VisionPsy-Nano. It is capable of analyzing images, documents, and charts, as well as answering questions without sending the source materials to remote systems. The ability of this model to process input and output on mobile phones marks a significant advancement in AI technology.