Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Marco Search Agent for Realistic and Challenging Agentic Search
| Date | Stars |
|---|---|
| 2026-07-31 | 324 |
| 2026-08-06 | 324 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<p align="center">
<img src="assets/logo_1.png" width="250" style="margin-bottom: 0.2;"/>
<p>
# 🍓 Marco DeepResearch: Towards Real‑World and Efficient Agents
[]()
[](https://opensource.org/licenses/Apache-2.0)
[](https://www.python.org/downloads/)
<img src="https://img.shields.io/github/issues/AIDC-AI/Marco-Search-Agent?color=red" alt="Open Issues">
<img src="https://img.shields.io/github/issues-closed/AIDC-AI/Marco-Search-Agent?color=green" alt="Closed Issues">
<img src="https://img.shields.io/github/stars/AIDC-AI/Marco-Search-Agent?color=yellow" alt="Stars">
<div align="center">
🍓 [_**Alibaba Group**_](https://www.qianwenai.com/) 🍓
📝 [**Marco DeepResearch Technical Report**](https://arxiv.org/abs/2603.28376) | 📝 [**HSCodeComp Paper**](https://arxiv.org/abs/2510.19631) | 📝 [**DeepWideSearch Paper**](https://arxiv.org/abs/2510.20168) | 🤗 [**HSCodeComp Dataset**](https://huggingface.co/datasets/AIDC-AI/HSCodeComp) | 🤗 [**DeepWideSearch Dataset**](https://huggingface.co/datasets/AIDC-AI/DeepWideSearch) | 🤗 [**Marco-DeepResearch-8B Model**](https://huggingface.co/AIDC-AI/Marco-DeepResearch-8B)
[English](README.md) | [简体中文](README_zh.md)
</div>
---
## Introduction
**Marco DeepResearch** is a comprehensive initiative from Alibaba that advances real-world AI agent capabilities through challenging benchmarks and practical applications. Our work bridges the gap between AI agents and human experts by exposing and addressing critical limitations in domain-specific reasoning, hierarchical rule application, and large-scale information seeking.
<div align="center">
<img src="assets/Timeline_2.png" alt="Marco DeepResearch Roadmap" width="800">
</div>
### 🎯 Key Achievements
We introduce a suite of benchmarks, frameworks and models that evaluate and advance agents across fundamental dimensions essential for real-world deployment:
- **🏆 HSCodeComp**: Tests hierarchical rule application with **95.0% human performance** vs. **46.8% best AI** (SmolAgent + GPT-5 VLM)
- **🏆 DeepWideSearch**: Challenges deep-and-wide information seeking with **414 avg. information units** and **4.21 avg. reasoning depth**
- **🏆 Table-as-Search**: Production-ready hierarchical multi-agent framework demonstrating the **"scissor gap effect"** on challenging benchmarks
- **🏆 UMEM**: Self-evolving memory system that avoids the "rote memorization trap" through joint optimization of memory extraction and management
These benchmarks and frameworks reveal and address fundamental gaps in current AI systems for:
- Complex hierarchical decision-making in vertical domains (tariff, legal, medical, taxation)
- Simultaneous wide-scale exploration and deep multi-hop reasoning
- Structured information organization and synthesis
- Generalizable long-term memory that evolves without overfitting
---
## 🔥 News & Updates
* **[2026-05]** 🎉 Release **[Marco-DeepResearch-8B Model](https://huggingface.co/AIDC-AI/Marco-DeepResearch-8B)** and **[Marco-DeepResearch Agent](https://github.com/AIDC-AI/Marco-DeepResearch/blob/main/Marco-DeepResearch-Family/Marco-Agent-DeepResearch/README.md)**
* **[2026-05]** 🎉 **UMEM (Unified Memory Extraction and Management)** has been accepted by ICML 2026
* **[2026-04]** 🎉 Table-as-Search has been accepted by ACL 2026 Findings
* **[2026-04]** 🎉 HSCodeComp has been accepted by ACL 2026 Main-Oral (10: Top 5% of accepted papers, seminal paper)
* **[2026-03]** 🎉 Release **[Marco DeepResearch Technical Report](https://arxiv.org/abs/2603.28376)**
* **[2026-02]** 🎉 Released **UMEM (Unified Memory Extraction and Management)** - a self-evolving memory framework that jointly optimizes extraction and management for generalizable agent memory.
* **[2026-02]** 🎉 Released **Table-as-Search** - a structured planning framework for Excerpt of 18,300 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:547ebaef9de27b38, llm:Repository name and description: 'Marco Search Agent for Realistic and Challenging Agentic Search' — suggests an agent/search agent project focused on agentic search research (agent frameworks/agents). Language: Python.
matched fp:547ebaef9de27b38, llm:Repository name and description: 'Marco Search Agent for Realistic and Challenging Agentic Search' — suggests an agent/search agent project focused on agentic search research (agent frameworks/agents). Language: Python.
matched fp:547ebaef9de27b38, llm:Repository name and description: 'Marco Search Agent for Realistic and Challenging Agentic Search' — suggests an agent/search agent project focused on agentic search research (agent frameworks/agents). Language: Python.