Tongyi DeepResearch - Ali Tongyi Open Source Deep Research Intelligence Body

What is Generali DeepResearch?

Tongyi DeepResearch (Tongyi DeepResearch) is an open source intelligent body launched by Alibaba, designed for deep information retrieval and complex task reasoning, with 30 billion parameters, supporting multiple reasoning modes, including ReAct mode and depth mode, the latter of which significantly improves complex reasoning capabilities through an iterative research paradigm. Tongyi DeepResearch uses a full-process synthetic data solution that can automatically generate high-quality datasets without human intervention, breaking through the upper limit of capabilities. The Tongyi DeepResearch training process includes continuous pre-training, supervised fine-tuning and reinforcement learning, forming a complete end-to-end training link. Tongyi DeepResearch has been successfully applied in many fields, such as the AI Travel Agent of Gaode Map and "Tongyi Farrer" in the field of law, which has demonstrated its strong practicality and value.

Features of Tongyi DeepResearch

• in-depth information retrieval: Supporting the processing of long-cycle, multi-step complex information retrieval tasks, suitable for academic research, market analysis and other scenarios.
• multimodal reasoning: Supports the ReAct model and the depth model, the latter of which enhances complex reasoning through the IterResearch paradigm.
• Automatic data generation: Based on a full-process synthetic data solution, high-quality datasets can be generated without human intervention, breaking through the upper limit of the intelligence's capabilities.
• Enhanced Learning Optimization: Ensure that the behavior of intelligences is consistent with higher-order goals through customized reinforcement learning algorithms to improve the model's adaptability and stability in dynamic environments.
• application-enabledThe company has been successfully applied in many practical scenarios, such as the AI Travel Agent of Gaode Maps and the "Tongyi Farrer" in the legal field, which has demonstrated its strong practicality and value.

Core Strengths of Tongyi DeepResearch

• Powerful reasoning: Through the IterResearch paradigm (IterResearch), complex tasks are decomposed into multiple research rounds, gradually streamlining the workspace and dynamically reconfiguring the workspace to effectively improve the quality of complex reasoning and decision-making.
• Full Process Synthesis Data: It automatically generates high-quality datasets without human intervention, supports a complete training chain from pre-training to fine-tuning to reinforcement learning, and breaks through the upper limit of the intelligence's capability.
• End-to-end intensive learning: A customized reinforcement learning algorithm is used to ensure that the behavior of the intelligences is consistent with the higher-order goals, and to improve the model's adaptability and stability in dynamic environments, while avoiding instability and premature convergence problems during the training process.
• Efficient data management: Optimize data in real time guided by training dynamics and dynamically adjust the training set through fully automated data synthesis and data funnels to ensure training stability and performance improvement.

What is Tongyi DeepResearch's official website?

• Project website:: https://tongyi-agent.github.io/blog/introducing-tongyi-deep-research/
• Github repository:: https://github.com/Alibaba-NLP/DeepResearch
• HuggingFace Model Library:: https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B
• arXiv Technical Paper::https://arxiv.org/pdf/2509.13310

Who Tongyi DeepResearch is for?

• (scientific) researcher: To help researchers efficiently complete complex academic research tasks and improve research efficiency.
• Corporate Analyst: Provide competitor analysis, industry trend reports, etc., to help companies develop accurate market strategies, is the right hand of business analysts.
• legal practitionerIn the legal field, applications such as "Tongyi Farrer", which automatically searches for laws, cases and decisions, provide powerful productivity tools for legal practitioners.
• data scientist: Provides data scientists with powerful tools for generating high-quality datasets and optimizing model training through full-process synthetic data solutions and end-to-end reinforcement learning.