1. Increase support for visual models and multimodal models; 2. The application and workflow can input various multimodal files such as images, documents, audio, and videos.
