The state of agents that read the internet for you

What it does This repository is a curated awesome-list tracking the rapidly expanding ecosystem of agentic deep research. It catalogs commercial offerings—Gemini Deep Research, OpenAI’s platform, Perplexity, Grok, Copilot Researcher, and others—alongside open-source alternatives like DeerFlow, GPT-Researcher, and Auto-Deep-Research, plus recent papers and benchmarks. The maintainers also link to their own ACL 2026 position paper framing the shift from traditional web search to reasoning-driven search agents.

The interesting bit The repository reads like a field map for a land grab: every major lab and startup now has a “deep research” product, and this list attempts to keep the taxonomy straight. Its chronological paper index breaks down base models, optimization recipes such as GRPO and DPO, search engines, and evaluation datasets, making it more useful for comparing training approaches than reading press releases.

Key highlights

• Surveys 11 commercial products across Google, OpenAI, Anthropic, Microsoft, xAI, ByteDance, Alibaba, Moonshot, and others
• Indexes 17 open-source implementations with live GitHub star badges and one-line descriptors
• Research table details training optimizations (SFT, DPO, GRPO, REINFORCE++), agent architectures, and benchmark coverage
• Tied to an ACL 2026 position paper on incentivizing search with reasoning agents
• Points to a companion Awesome-RAG-Reasoning list for broader retrieval-and-reasoning context

Caveats

• The README leans heavily on emoji hype markers ("🔥🔥🔥", “🚀🚀🚀”) that dilute the signal
• As a curated index rather than a framework, it provides no executable code or reproducible experiments itself
• Several paper entries in the table lack linked code repositories or complete evaluation details

Verdict Worth bookmarking if you’re building or benchmarking research agents and need a panoramic view of who is doing what; skip it if you want a single reusable codebase, because this is strictly a reading list.