Timeline

OpenAI introduces Canvas, a collaborative writing and coding interface

The beta interface opened a separate editing pane where users could highlight text for targeted rewrites or annotate code, rather than regenerating whole chat replies.

  • Models & capabilities
  • Notable

OpenAI introduced canvas, a new ChatGPT interface for writing and coding tasks that opens a separate editing window alongside the chat, rather than confining work to a linear back-and-forth of full message replies. In the beta, available to ChatGPT Plus and Team users by selecting “GPT-4o with canvas” in the model picker, the model could open a canvas automatically when it judged a task would benefit from a dedicated workspace, such as a long piece of writing or a coding project.

Inside canvas, users could highlight a specific sentence or code block and ask for a targeted edit — instructions like “make this sound friendlier” or requests to adjust length via a slider — without regenerating the entire document, and for code the tool could insert inline comments explaining what a block did. OpenAI said Enterprise and Edu users would gain access the following week, with a wider rollout to free users planned once the beta concluded.

Canvas moved ChatGPT’s interface away from pure conversational turn-taking toward direct manipulation of a persistent document, a pattern already familiar from Anthropic’s Artifacts feature launched earlier in 2024. The shift reflected a broader move across chat products that quarter, treating the model less as a respondent to be re-prompted and more as a collaborator editing a shared workspace — a step toward the agentic, task-completing interfaces labs pursued through the following year.