# Microsoft releases Mage-VL-Cowork 4B hybrid computer-use model

Digest AI · Generative AI & Models · published 2026-10-31T10:02:47Z

Canonical: https://digestai.news/story/microsoft-releases-mage-vl-cowork-4b-hybrid-computer-use-model

## Summary

Microsoft unveiled Mage‑VL‑Cowork, a 4 billion‑parameter vision‑language model that can both interpret GUI screenshots and execute terminal commands without an external orchestrator. Trained from an intermediate Mage‑VL checkpoint through a five‑stage specialize‑then‑unify curriculum, the model reaches 66.0% success on the OSWorld‑Verified benchmark, the highest score among 4 B models, and scores 87.5% on pure OS tasks and 62.7% on multi‑application workflows.

The training pipeline, called Mage‑CUA‑Forge, generated 293,806 tasks covering 6,915 atomic skills, cutting recurring per‑task generation cost by 82.6% (from $0.906 to $0.158) and token usage by 93.3% (1,276 K → 85 K tokens). Online hybrid reinforcement learning added +10.16 pp accuracy while reducing mean model calls by 13.0% and output tokens by 25.4%. The model is released with a Docker container and SGLang server support, using a custom multimodal architecture (llavaonevision2) and a 32,768‑token context window.

## Key points

- Mage‑VL‑Cowork achieves 66.0% OSWorld‑Verified score, best among 4 B models
- Training pipeline generated 293,806 tasks, lowering per‑task cost to $0.158 (‑82.6%)
- Online hybrid RL improves accuracy by +10.16 pp and cuts token usage by 25.4%

## Why it matters

A compact 4 B multimodal model that reliably handles full‑desktop and terminal tasks narrows the gap between large‑scale agents and on‑device AI, enabling more efficient, privacy‑preserving automation.

## Sources

1. [microsoft/Mage-VL-Cowork · Hugging Face](https://huggingface.co/microsoft/Mage-VL-Cowork) (huggingface.co, 2026-10-31, primary source)

## Cite

Digest AI, "Microsoft releases Mage-VL-Cowork 4B hybrid computer-use model", 31 October 2026, https://digestai.news/story/microsoft-releases-mage-vl-cowork-4b-hybrid-computer-use-model

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/microsoft-releases-mage-vl-cowork-4b-hybrid-computer-use-model.json