Mini toolsFara1.5 27B
By Microsoft
Fara1.5-27B is a vision-only computer use agent supervised fine-tuned from Qwen3.5-27B by Microsoft Research AI Frontiers. It perceives the browser purely through screenshots, without DOM or accessibility tree access, and predicts pixel-grounded next actions such as clicks, drags, typing, scrolling, and navigation. It supports a 262K token context for tracking multi-screenshot trajectories with full action history. The model uses a critical-points safety framework that pauses execution before entering personal information, making payments, submitting forms, sending messages, or other irreversible actions, requiring explicit user authorization. Intended for automating repetitive web tasks like form filling, shopping, travel booking, and information seeking, and can serve as a grounding model for other agents needing pixel-accurate action prediction. Recommended deployment is inside MagenticLite, adding sandboxing, allow-lists, and monitoring.
Overview
Fara1.5-27B is a 27B-parameter multimodal computer use agent for web browsers. It observes browser screenshots and emits structured tool calls (click, type, scroll, visit URL, web search) to complete end-to-end tasks like filling forms, shopping, and booking travel, pausing for user confirmation at safety-critical points like payments or submissions.
About Microsoft
Microsoft is a technology company that offers a wide range of software, cloud computing services, hardware, and artificial intelligence solutions.
Industry:
Technology, Information and Internet
Company Size:
228000
Location:
Redmond, Washington, US
Website:
microsoft.com
Last updated: July 23, 2026
