Qwen DianJin is an open-source platform by Alibaba Cloud that provides a suite of financial LLMs and LMMs, along with integrated AI technologies, to facilitate the development of financial applications. It targets financial business developers, offering tools and APIs for tasks like research report summarization, financial news analysis, and customer service automation, aiming to drive innovation in the financial sector.

How It Works

The platform integrates various AI technologies, including LLMs and LMMs, to offer standardized API capabilities for financial scenarios. Key features include Document Q&A with optimized parsing and recall strategies for financial knowledge bases, Metrics Q&A for understanding and plotting financial metrics, and a Multi-Agent System for flexible orchestration of capabilities. This approach aims to provide a comprehensive and adaptable financial intelligence solution.

Quick Start & Requirements

• Models: DianJin-R1-7B, DianJin-R1-13B, DianJin-R1-32B, Fin-PRM, DianJin-OCR-R1.
• Datasets: DianJin-R1-Data, CFLUE, M³FinMeeting.
• Resources: Links to HuggingFace and ModelScope for models and data. Technical reports and papers are available for detailed information.

Highlighted Details

• Features specialized models like Fin-PRM for financial reasoning and DianJin-OCR-R1 for enhanced OCR capabilities via a vision-language model.
• Includes the CFLUE dataset, benchmarked in an ACL-2024 paper, and the M³FinMeeting dataset accepted by ACL-2025.
• Offers a multi-agent system for custom financial application development.

Maintenance & Community

• Contact via email (CFLUE@alibabacloud.com) or DingTalk group for support.
• Active research and development with recent publications and open-source releases.

Licensing & Compatibility

• The README does not explicitly state the license type or compatibility for commercial use. A disclaimer notes that users are responsible for assessing and assuming risks associated with using the models and data.

Limitations & Caveats

• The platform's licensing and commercial use terms are not clearly defined in the provided README. Users are advised to exercise caution and independently verify model outputs.