v.2.0.0

microsoft/OmniParserv.2.0.0Feb 13, 2025by yadong-lu

AI Summary

Major version 2.0.0 release featuring a larger and cleaner dataset, 60% latency improvement over V1, strong ScreenSpot Pro performance, and OmniTool for controlling Windows 11 VMs with support for multiple LLM providers.

Key Highlights

  • Larger and cleaner icon caption + grounding dataset
  • 60% improvement in latency compared to V1
  • 39.6 average accuracy on ScreenSpot Pro
  • OmniTool for Windows 11 VM control
  • Support for multiple LLM providers (OpenAI, DeepSeek, Qwen, Anthropic)

New Features

  • OmniTool for Windows 11 VM control
  • Support for OpenAI (4o/o1/o3-mini)
  • Support for DeepSeek (R1)
  • Support for Qwen (2.5VL)
  • Support for Anthropic Computer Use

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.