123 likesNew · 3d old
Microsoft releases 4B VLM that clicks and codes together
What is Mage-VL-Cowork?
Mage-VL-Cowork is a 4-billion-parameter vision-language model that can interpret screenshots, click through a GUI, and run terminal commands using a single policy. Trained on 293,806 synthetic tasks, it reaches 66 % success on the OSWorld benchmark, the best score for models of its size.
Who it is for
Developers of on-device assistants or automation pipelines can use it to prototype agents that both navigate graphical interfaces and execute shell scripts without stitching together separate models.
How to try it
Visit the Hugging Face model page and run the provided inference demo notebook.