OpenAI has introduced Canvas as a new interface to work with ChatGPT on writing and coding projects. Built with GPT-4o, Canvas is in early beta and can be manually selected in the model picker. The product opens a separate window beside the chat window, providing a workspace for writing and coding projects.

Canvas has been rolled out to ChatGPT Plus and Team users globally and will be accessible to Enterprise and Edu users this week. As stated in an official update, OpenAI plans to make it available to all ChatGPT-free users when it is out of beta.

The company stated that with Canvas, ChatGPT can better understand the context of what one wants to accomplish. It also acts as a copy editor or code reviewer to provide inline feedback and suggestions on the project, while the project is controlled by the user.

Also, canvas opens automatically if ChatGPT detects a scenario in which it could be used and when the user includes “use canvas” in the prompt. OpenAI has also affirmed training GPT-4o to collaborate as a creative partner. The model knows when to open a canvatrain required edits or rewrite entirely.

Moreover, the company has mentioned using novel synthetic data generation techniques to post-train the model for its core behaviors. This approach has proven beneficial in rapidly addressing writing quality and new user interactions without leaning on human-generated data.