# OmniParser New free and open source tool by Microsoft that helps LLMs better understand UIs by tokenizing UI screenshots into structured elements interpretable by LLMs. Will help create more powerful vision-based GUI AI agents. ## References - Announcement: https://www.microsoft.com/en-us/research/articles/omniparser-v2-turning-any-llm-into-a-computer-use-agent/ - Source code: https://github.com/microsoft/OmniParser