Many of today’s most powerful AI models are multimodal, capable of understanding context from both text and image inputs. To fully leverage these models, developers need a way to easily send image data as part of their API calls.

To address this need, the CallAIModel action has been updated to support sending images to AI models. This new functionality introduces a breaking change to the structure of the Content input parameter. Previously, this parameter was a simple Text field. It has now been replaced with a new Content structure that can hold either a ContentURL (Text) or ContentBinary (Binary Data) field. This new structure provides the flexibility to send image data in various formats, ensuring that your applications can interact with a wide range of multimodal AI models. This change enables developers to build more advanced AI applications that can see and interpret the world, empowering new and exciting user experiences.