# Beijing Academy of Artificial Intelligence

Beijing Academy of Artificial Intelligence (BAAI), also known as Zhiyuan Institute, is a Chinese non-profit AI research laboratory focused on fundamental research, open-source infrastructure, and large-scale models like WuDao and Emu3.

The Beijing Academy of Artificial Intelligence (BAAI), also known as the Zhiyuan Institute, is a Chinese non-profit research laboratory dedicated to advancing artificial intelligence. Founded as a collaborative hub, BAAI brings together leading AI companies, universities, and research institutes from China and abroad. The academy focuses on long-term fundamental research, fostering top talent, and promoting cooperation between academia and industry. It is considered one of the pre-eminent AI research institutes in China, frequently releasing new models and open-source code to support the broader AI community.

BAAI organizes an annual international conference that convenes AI experts, industry leaders, and international researchers to discuss challenges and the future direction of the field. As of 2023, the academy's research priorities center on large pre-trained models, including large language models, and open-source AI infrastructure.

## WuDao

WuDao (Chinese: 悟道; pinyin: wùdào) is a large multimodal pre-trained language model developed by BAAI. WuDao 2.0, announced on 31 May 2022, was compared to [GPT-3](https://www.wikiprompt.org/wiki/openai) at the time, but with a significantly larger scale: WuDao 2.0 contains 1.75 trillion parameters, compared to GPT-3's 175 billion, making it the largest pre-trained model in the world at that time. The model was trained on 4.9 terabytes of images and text, including 1.2 terabytes of Chinese text and 1.2 terabytes of English text.

The chairman of BAAI stated that WuDao was an attempt to "create the biggest, most powerful AI model possible," though direct comparisons based solely on parameter count do not necessarily correlate with model quality. WuDao demonstrated capabilities in natural language processing, image recognition, and generation of both text and images. It could write essays, poems, and couplets in traditional Chinese, generate text from static images, and produce nearly photorealistic images from natural language descriptions. The model also powered virtual chat agents and showed an ability to predict the 3D structures of proteins, similar to [AlphaFold](https://www.wikiprompt.org/wiki/google-deepmind).

## FlagAI and Jiuding

FlagAI is an open-source, extensible toolkit for large-scale model training and inference. Its goal is to support training, fine-tuning, and deployment of large-scale models on various downstream tasks with multi-modality. The open repository includes all source code as well as several pre-trained large models. FlagAI is an approved incubation project at the sandbox level of the Linux Foundation.

Jiuding is an AI-computing platform that focuses on supporting AI innovation. As of September 2022, it provides 1000P computation capacity with 400 Gbit/s high-speed interconnection per server and supports AI chipsets of different architectures. The platform also includes code compilers for the various AI architectures.

## Worm Models

MetaWorm is a computational model of the Caenorhabditis elegans (C. elegans) nematode, simulating the worm's nervous system along with a "digital body" simulation in real-time. MetaWorm 1.0 exhibits behaviors that parallel C. elegans in the real world.

BAAIWorm is an integrative data-driven model of C. elegans, consisting of two submodels: the brain model and the body-environment model. These projects reflect BAAI's interest in computational biology and embodied intelligence.

## Emu3 and BGE

Emu3 is a suite of multimodal AI models trained solely with next-token prediction on tokenized images, text, and videos. This approach aligns with recent trends in [generative AI](https://www.wikiprompt.org/wiki/generative-ai) and unified modeling across modalities.

BGE, which stands for BAAI General Embedding, is a series of embedding models developed and published by BAAI. These models are widely used for tasks such as semantic search and retrieval, contributing to the academy's open-source ecosystem.

## U.S. Sanctions

In March 2025, the U.S. Commerce Department added BAAI to the Entity List for allegedly developing technology for military purposes. This designation restricts certain exports and collaborations with U.S. entities, reflecting geopolitical tensions in the AI sector.

## See Also

- [Artificial intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence)
- [Machine learning](https://www.wikiprompt.org/wiki/machine-learning)
- [Large language model](https://www.wikiprompt.org/wiki/large-language-model)
- [Transformer](https://www.wikiprompt.org/wiki/transformer)

## External links

- [Wikipedia: Beijing Academy of Artificial Intelligence](https://en.wikipedia.org/wiki/Beijing_Academy_of_Artificial_Intelligence)

---
Source: https://www.wikiprompt.org/wiki/beijing-academy-of-artificial-intelligence
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-14T04:22:00.403947+00:00