DavidZWZ/Awesome-Deep-Research

Awesome-Deep-Research is a curated, constantly-updated guide to "Agentic Deep Research" — a field where AI systems act as autonomous research agents, searching the web, gathering evidence, reasoning over it, and producing thorough answers or reports, rather than just returning a ranked list of links like a standard search engine.

What's inside

The repo is a single, structured list of resources organized into four sections:

1. Industry-leading products — 12 commercial tools and platforms, each with a name, a link, and a release date. These are the big players: Google's Gemini Deep Research, OpenAI's Deep Research, Perplexity Deep Research, xAI's Grok Agents, Microsoft's Copilot Researcher, Anthropic's Research, Manus, ByteDance's DeerFlow, Alibaba's Qwen Deep Research, Moonshot's Kimi-Researcher, plus two newer search/API services (Not Human Search, SearchHive).

2. Open-source implementations — 19 GitHub repositories you can actually run yourself. These include full research frameworks, lightweight toolkits, and AI-agent builders — e.g., gpt-researcher (autonomous agent for research), DeerFlow (ByteDance's framework), OpenManus (general AI-agent builder), and Local Deep Research (runs on your own machine, pulls from web/arXiv/PubMed/private documents, uses 20+ research strategies).

3. Latest research papers — a detailed chronological table (newest at top) of academic papers in the field. Each row lists the paper title (linked to arXiv), date, base model used, optimization technique (SFT, RL, etc.), search engine, agent architecture (single vs. multi-agent), training dataset, and evaluation dataset. The README notes the papers cover topics like agentic frameworks, search-enhanced reasoning models, web agents, retrieval-augmented generation, and multimodal deep research.

4. Evaluation benchmarks and practical applications — the README's table of contents mentions this section, though the content shown is just the section header at this point.

Who it's for

Researchers, developers, and enthusiasts interested in the intersection of artificial intelligence and autonomous agents — specifically, building systems that can perform deep, multi-step research tasks.

Notable context

• The repo also links to the maintainers' own position paper on arXiv: "From Web Search towards Agentic Deep Research" — a discussion of why research is shifting from simple keyword search to reasoning-driven agents, and where the field is heading.
• It cross-references a sister repo, Awesome-RAG-Reasoning, for those interested in the broader topic of retrieval-augmented generation plus reasoning.
• Badges indicate the project is maintained, contributions are welcome, and the README explicitly encourages contributions.

Bottom line

It's a link-and-annotation hub — not a tool itself, but a directory of the best products, open-source code, and academic work in the emerging field of AI-powered deep research, kept current with dates and structured detail (models, datasets, architectures) to help you compare and choose what to use or study.