Skip to content

Ground-shaking Project in Q1 2025 on a Cost-Effective Personal AI Assistant: "Auto Deep Research" Developed by Prof Chao Huang's Team as an Alternative to OpenAI's Deep Research!

Professor Chao HUANG

Assistant Professor
HKU IDS / CDS

A research team led by Professor Chao Huang, Assistant Professor at HKU’s Institute of Data Science (IDS) and School of Computing and Data Science, has developed “Auto Deep Research” – an innovative open-source personal AI assistant.

As a leading open-source solution, the system demonstrates performance comparable to OpenAI’s Deep Research, consistently maintaining top positions on the GAIA Benchmark for general AI assistants. Launched in Spring 2025 as a cost-effective alternative to OpenAI’s Deep Research, the project has garnered significant attention from mainstream tech media and tech giants.

 

Quick Links:

📦 Auto Deep Research: https://github.com/HKUDS/Auto-Deep-Research

⚙️ Powered by AutoAgent Framework: https://github.com/HKUDS/AutoAgent

🎓 Further Reading: https://arxiv.org/abs/2502.05957

(The first author of this article is a 2nd Year PhD candidate at HKU IDS, Mr. Jiabin Tang, supervised by Professor Chao Huang) 

Key Features of Auto Deep Research 

  1. Cost-Effectiveness & Open Source 
    While OpenAI’s Deep Research’s high costs in subscriptions of US$200 per month have been a concern to many users, Prof Huang’s research project on Auto Deep Research offers a cost-efficient alternative, allowing one to use his/her own API keys with a pay-as-you-go model while maintaining premium performance. As an open-source solution, it ensures transparency and community-driven improvements. 

  2. High Performance Excellence 
    As a leading open-source solution, the system derived by Prof Huang’s team demonstrates performance comparable to OpenAI’s Deep Research, consistently maintaining top positions on GAIA Benchmark as general AI Assistant. 

  3. Ensuring a Seamless Out-of-the-Box User Experience 
    Auto Deep Research renders a completely different user experience – it is as simple as a one-click deployment system. It eliminates complex setup procedures, allowing one to launch advanced research capabilities instantly with a single command.  

  4. Universal Compatibility and Flexibility with Other LLMs 
    5. The system supports integration with a wide range of Large Language Models (including OpenAI, Anthropic, DeepSeek, and others), offering flexible interaction modes with both function-calling and non-function-calling capabilities, along with comprehensive file handling support. 

A big round of applause to Prof Huang and his team for this ready-to-use, affordable AI tool. For the full biography of Prof Huang, please visit: https://datascience.hku.hk/people/chao-huang/  

Prof Huang’s work has been featured in key pages for disseminating sci-tech news. If you are interested, please browse the article here.