BAAI (Beijing Academy of Artificial Intelligence)

Beijing Academy of Artificial Intelligence (BAAI), commonly known in Chinese as 智源研究院, is a non-profit artificial intelligence research institute based in Beijing, China. It was established in 2018 as a collaborative initiative bringing together Chinese universities, research institutes and technology companies to advance fundamental and applied AI research. The institute operates as a research platform rather than a conventional commercial vendor, funding and coordinating research programs across machine learning, natural language processing, computer vision and large-scale models. BAAI is best known publicly for its open research initiatives and open-source model releases. Its flagship program, often referred to as the "Wudao" (悟道) large-model project, positioned the institute among the early Chinese organizations working on very large pre-trained language models. The institute has also released open model families and datasets intended for the research community, and it runs academic exchange activities, talent programs and benchmark efforts. As an entity, BAAI functions as a research brand and convening platform: it partners with universities and companies, hosts conferences and workshops, and publishes research outputs. Its positioning is that of an open, non-profit research institute supporting China's broader AI ecosystem rather than a consumer-facing product company. Note: the supplied evidence snippets concern an unrelated news story about Truth Social and do not describe this institute; the dossier above relies on well-established public facts about BAAI and should be verified against the institute's official materials.

The Beijing Academy of Artificial Intelligence (BAAI) was established in Beijing in 2018 as a non-profit research institute created through collaboration among Chinese universities, research organizations and technology companies. Its founding model was unusual: rather than a single corporate owner, it was designed as a shared platform to coordinate AI research capacity across institutions. In its early years BAAI built out research programs and fellowship and talent initiatives, and it became known for organizing academic exchange in the Chinese AI community. It gained wider public recognition with the launch of its large-scale pre-trained model initiative, commonly called "Wudao" (悟道), which placed it among the early Chinese efforts to build very large language models. The institute subsequently emphasized open-source releases of models, datasets and tools, positioning itself as an open research brand. BAAI has continued to operate as a convening body, hosting conferences and workshops and partnering with universities and companies on model development and evaluation. Its public identity rests on open research, talent cultivation and large-model work rather than on commercial product sales. Note: the supplied evidence snippets concern an unrelated news story and contain no information about this institute; the history above relies on well-established public facts and should be verified against official BAAI materials.

• 2018

  Beijing Academy of Artificial Intelligence (BAAI) established in Beijing as a non-profit AI research institute.

• Launch of the "Wudao" (悟道) large-scale pre-trained model initiative.

• Release of open-source models, datasets and benchmarks for the research community.

Non-profit, open AI research institute and convening platform supporting China's AI ecosystem.

• Wudao (悟道) large-model program

• Open-source model releases

• Research datasets and benchmarks

• 2026-08-13

  BAAI FlagOS completes Day0 multi-chip adaptation for Alibaba's Qwen3.8-2.4T-A95B[2]智源FlagOS完成阿里Qwen3.8-2.4T-A95B的Day0多芯片适配163.com

  Reported: 2026-08-13

  On August 13, the Beijing Academy of Artificial Intelligence (BAAI) announced that its Zhongzhi FlagOS community completed Day0 multi-chip adaptation for Alibaba's open-source ultra-large MoE model Qwen3.8-2.4T-A95B, with adaptation, precision alignment and deployment validation done on nine AI chips including T-Head, Nvidia, Moore Threads, Huawei Ascend, MetaX, Kunlunxin, Hygon, Qingwei and Suiyuan, offering BF16, FP8 and INT8 precision versions.

1. 1Beijing Academy of Artificial Intelligencebaai.ac.cn
2. 2智源FlagOS完成阿里Qwen3.8-2.4T-A95B的Day0多芯片适配163.com