AI BriefingMage-VL-Cowork: Microsoft Mage-VL-Cowork: A Hybrid Agent Navigating Between GUI and Terminal
microsoft/Mage-VL-Cowork
About the project
Mage-VL-Cowork, released by Microsoft, is a computer-use agent based on a vision-language model (VLM). It features a hybrid agent architecture that understands and operates both the GUI and terminal environments on the screen, going beyond simple text generation.
This model is designed to perform the roles of a GUI agent, coding agent, and terminal agent within a single framework. Trained through reinforcement learning (RL), it demonstrates verified performance on real-world computer task benchmarks such as OSWorld and WebVoyager.
Unlike existing single-domain specialized models, it handles a wide range of tasks from GUI operations like mouse clicks to code writing and terminal command execution. It is based on the LLaVA-OneVision2 architecture and is distributed under the Apache 2.0 license, making it available for use in the open-source ecosystem.
It is suitable for developers building AI agents for computer-use (CUA) automation or handling complex multi-step tasks. You can run inference directly via the Transformers library or serve it as an API through Hugging Face endpoints.
microsoft/Mage-VL-Cowork
The original page has no description.
image-text-to-text
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.