Shandu - Open source OpenAI DeepResearch alternative

Release : Shandu - OpenAI DeepResearch alternative, An AI-driven research system that performs comprehensive, iterative research on any topic using multiple search engines and LLMs. Key Features Iterative Research: Recursively explores topics through multiple search engines with thematic organization Ethical Web Scraping: Respects robots.txt rules and implements caching to minimize server impact Comprehensive Reports: Generates 7000+ word detailed, well-structured markdown research reports Configurable Parameters: Fine-tune research depth and breadth to suit your specific needs Source Evaluation: Automatically assesses reliability and credibility of information sources Parallel Processing: Optimized with concurrent operations for more efficient execution Lightweight Search: Quick AI-powered search alternative with the aisearch command Target Audience People who don't want to pay 20$ for 10 queries a month to OpenAI or 200$ (for a Pro plan). Comparison It usually finishes report on any topic before OpenAI with usually much more sources. I didn't still check money it takes but it's usually like cents. You can also use any model, even locally run ones (but report will be worse, of course). You can run it with cheap api like DeepSeek. How it works Initial Setup Takes user query and research parameters (breadth & depth) Generates follow-up questions to understand research needs better Deep Research Process Generates multiple SERP queries based on research goals Processes search results to extract key learnings Generates follow-up research directions Recursive Exploration If depth > 0, takes new research directions and continues exploration Each iteration builds on previous learnings Maintains context of research goals and findings Report Generation Compiles all findings into a comprehensive markdown report Includes all sources and references Organizes information in a clear, readable format Examples: Check examples/ folder. Typical report (note this is with model of 72b params, you can achive much better results with o3 mini/o3-mini-high , just put openrouter api base,key, and model name). https://github.com/jolovicdev/shandu/blob/main/examples/qwen72b-instruct-test_nsa_esa_cnsa.md GitHub repository: https://github.com/jolovicdev/shandu

Feb 27, 2025 - 18:33

Shandu - Open source OpenAI DeepResearch alternative

Release : Shandu - OpenAI DeepResearch alternative, An AI-driven research system that performs comprehensive, iterative research on any topic using multiple search engines and LLMs.

Key Features
Iterative Research: Recursively explores topics through multiple search engines with thematic organization

Ethical Web Scraping: Respects robots.txt rules and implements caching to minimize server impact

Comprehensive Reports: Generates 7000+ word detailed, well-structured markdown research reports

Configurable Parameters: Fine-tune research depth and breadth to suit your specific needs

Source Evaluation: Automatically assesses reliability and credibility of information sources

Parallel Processing: Optimized with concurrent operations for more efficient execution

Lightweight Search: Quick AI-powered search alternative with the aisearch command

Target Audience
People who don't want to pay 20$ for 10 queries a month to OpenAI or 200$ (for a Pro plan).

Comparison
It usually finishes report on any topic before OpenAI with usually much more sources. I didn't still check money it takes but it's usually like cents. You can also use any model, even locally run ones (but report will be worse, of course). You can run it with cheap api like DeepSeek.

How it works
Initial Setup

Takes user query and research parameters (breadth & depth)

Generates follow-up questions to understand research needs better

Deep Research Process

Generates multiple SERP queries based on research goals

Processes search results to extract key learnings

Generates follow-up research directions

Recursive Exploration

If depth > 0, takes new research directions and continues exploration

Each iteration builds on previous learnings

Maintains context of research goals and findings

Report Generation

Compiles all findings into a comprehensive markdown report

Includes all sources and references

Organizes information in a clear, readable format

Examples: Check examples/ folder.

Typical report (note this is with model of 72b params, you can achive much better results with o3 mini/o3-mini-high , just put openrouter api base,key, and model name). https://github.com/jolovicdev/shandu/blob/main/examples/qwen72b-instruct-test_nsa_esa_cnsa.md

GitHub repository: https://github.com/jolovicdev/shandu