Microsoft launched a Visual ChatGPT — a combination of ChatGPT and a series of Visual Foundation Models

Visual ChatGPT enables users to send, receive and edit images during chatting.

In other words, Visual ChatGPT supports talking, drawing and editing. This is the next advanced stage in development of large language models.

According to Miscrosoft team, the system combines different Visual Foundation Models and activates more options of interaction with ChatGPT.

Visual ChatGPT enables to:

1. send and receive not only text lines but also images,
2. provide complex visual questions or visual editing instructions that require the collaboration of multiple AI models with multi-steps,
3. provide feedback and ask for amended results.

Instead of training a new multi-modal  ChatGPT from scratch the developers built Visual  ChatGPT directly based on ChatGPT and incorporated a variety of Visual Foundation Models, such as Visual Transformers or Stable Diffusion.