Open WebUI v0.13.0: Tabs, Terminal and Browser Inside the Chat

Open WebUI v0.13.0 has been released — a major interface update for running neural networks. The main change: a tabbed workspace that combines chat, files, terminal, artifacts, previews, voice calls and comments. Details are in the release notes on GitHub.

What it means

For those running models on their own hardware, the update adds two new modes of operation. The Desktop tab provides access to the desktop of a connected Open Terminal: you can control the mouse, keyboard, clipboard and sound, if the connection supports it. A model capable of calling desktop tools can perform computer use tasks, and the user can intervene manually. The Browser tab adds a browser next to the chat: navigation, multiple tabs, mobile viewport. In Chrome mode on a compatible Open Terminal, the model and the user get a shared browser for automation — navigating websites, working with forms and testing web applications that the model creates. Proxy mode remains available for preview. Both modes require an Open Terminal connection with desktop or browser support enabled. The developer does not specify which models are compatible with the desktop and browser tools, and does not provide hardware requirements. The update targets scenarios where chat, code and result verification happen in a single window: data analysis, document generation, and web application development and testing. New releases of tools for running neural networks locally can be followed in the CPU3D news feed.