Project Title

DeepResearch — Leading Open-source Deep Research Agent for Long-Horizon Information Seeking

Overview

DeepResearch, also known as Tongyi DeepResearch, is an open-source deep research agent developed by Tongyi Lab. It features a 30.5 billion parameter large language model with 3.3 billion activated per token, designed for long-horizon, deep information-seeking tasks. The model demonstrates state-of-the-art performance across various agentic search benchmarks.

Key Features

• Fully automated synthetic data generation pipeline for agentic pre-training, supervised fine-tuning, and reinforcement learning.
• Large-scale continual pre-training on diverse, high-quality agentic interaction data.
• End-to-end reinforcement learning using a customized Group Relative Policy Optimization framework.
• Agent Inference Paradigm Compatibility for ReAct and IterResearch-based 'Heavy' mode.

Use Cases

• Researchers and developers using DeepResearch for complex information-seeking tasks requiring deep reasoning and understanding.
• Enterprises leveraging the model for advanced search and data analysis applications.
• Educational institutions using the model for teaching and research in natural language processing and artificial intelligence.

Advantages

• State-of-the-art performance in agentic search benchmarks.
• Scalable data synthesis pipeline for efficient model training and enhancement.
• Compatibility with different inference paradigms to balance between model performance and resource usage.

Limitations / Considerations

• The model's large size may require significant computational resources for training and inference.
• The complexity of the model may pose challenges for developers new to deep research agents.

Similar / Related Projects

• HuggingFace Transformers: A library of pre-trained models for natural language processing, offering a range of models but not specialized in deep research tasks like DeepResearch.
• OpenAI GPT: A series of large-scale language models that have shown impressive capabilities in various NLP tasks, but not specifically tailored for deep information-seeking as DeepResearch is.
• Stanford's Alpaca: A model that also focuses on long-form question-answering, but differs in its approach and training methodology compared to DeepResearch.

Basic Information

• GitHub: https://github.com/Alibaba-NLP/DeepResearch
• Stars: 11,237
• License: Unknown
• Last Commit: 2025-09-20

📊 Project Information

• Project Name: DeepResearch
• GitHub URL: https://github.com/Alibaba-NLP/DeepResearch
• Programming Language: Python
• ⭐ Stars: 11,237
• 🍴 Forks: 839
• 📅 Created: 2025-01-09
• 🔄 Last Updated: 2025-09-20

🏷️ Project Topics

Topics: [, ", a, g, e, n, t, ", ,, , ", a, l, i, b, a, b, a, ", ,, , ", a, r, t, i, f, i, c, i, a, l, -, i, n, t, e, l, l, i, g, e, n, c, e, ", ,, , ", d, e, e, p, -, r, e, s, e, a, r, c, h, ", ,, , ", d, e, e, p, r, e, s, e, a, r, c, h, ", ,, , ", i, n, f, o, r, m, a, t, i, o, n, -, s, e, e, k, i, n, g, ", ,, , ", l, l, m, ", ,, , ", t, o, n, g, y, i, ", ,, , ", w, e, b, -, a, g, e, n, t, ", ]

🔗 Related Resource Links

🌐 Related Websites

• [
• [
• [
• WebAgent
• 🤗 HuggingFace

This article is automatically generated by AI based on GitHub project information and README content analysis