Aria-UI

January 23, 2025 ยท View on GitHub

Overview

Aria-UI is a closed-source agent system developed by University of Hong Kong & Rhymes AI that combines multiple input modalities for computer interaction.

Key Features

  • Integration with GPT-4o
  • Multimodal approach
  • Joint development by HKU and Rhymes AI

Performance

OSWorld Results

  • Aria-UI w/ GPT-4o: 15.15%

Technical Details

  • Base Model: GPT-4o integration
  • Input: Multiple modalities
  • Focus: Computer interaction tasks

References