📱 PhoneBuddy-4B — Phone-Use GUI Agent

Give the agent a phone screenshot and a task instruction; it predicts the next action as a structured tool-call (click, type, scroll, open_app, …) with coordinates on a 0–1000 grid. The predicted tap location is drawn on the screenshot.

Model: PhoneBuddyAI/PhoneBuddy-4B · Project: phonebuddyai.github.io