v.2.0.0

svenstaro/genactv.2.0.0Feb 13, 2025by yadong-lu

AI Summary

A major version update introducing a new model (OmniParser-v2.0) featuring significant latency improvements and a unified tooling approach called OmniTool for controlling Windows 11 VMs.

Key Highlights

  • 60% improvement in latency compared to V1
  • 39.6 average accuracy on ScreenSpot Pro
  • OmniTool unifies the tooling (OmniParser + Vision Model)
  • Supports OpenAI (4o/o1/o3-mini), DeepSeek (R1), Qwen (2.5VL), and Anthropic Computer Use
  • Larger and cleaner set of icon caption + grounding dataset

New Features

  • OmniTool
  • Support for OpenAI, DeepSeek, Qwen, and Anthropic models
  • New model checkpoints (OmniParser-v2.0)
  • 60% latency improvement
  • ScreenSpot Pro accuracy of 39.6

Full Release Notes

# What's new in V2.0.0?
- Larger and cleaner set of icon caption + grounding dataset
- 60% improvement in latency compared to V1 [model checkpoints](https://huggingface.co/microsoft/OmniParser-v2.0)
- Strong performance: 39.6 average accuracy on [ScreenSpot Pro](https://github.com/likaixin2000/ScreenSpot-Pro-GUI-Grounding)
- Your agent only need one tool: OmniTool. Control a Windows 11 VM with OmniParser + your vision model of choice. OmniTool supports out of the box the following large language models - OpenAI (4o/o1/o3-mini), DeepSeek (R1), Qwen (2.5VL) or Anthropic Computer Use.