BAAI Innovation Center

BAAI Innovation Center

Profile

The Beijing Academy of Artificial Intelligence (abbreviated as "BAAI") is a non-profit new research and development institution established in November 2018. It is committed to leading AI innovation, creating the world's premier academic and technological innovation ecosystem, tackling the most profound challenges and pivotal issues, serving as a global hub for AI academic concepts, foundational theories, top-tier talent, corporate innovation, and policy development, and fostering the sustainable development of human beings, the environment and intelligence.

Tiejun Huang

Chairman, BAAI Professor, School of Computer Science, Peking University

Zhongyuan Wang

President, BAAI

Yonghua Lin

Vice President, BAAI Chief Engineer, BAAI

Milestones

Be the first to foresee the arrival of the big model era of AI

2024 12

BAAI Worm was published as cover story of Nature Computational Science

2024 10

Emu 3, the native multimodal world model, is officially released

2024 10

BGE, the top model of the monthly download ranking list of Hugging Face in October 2024

2024 06

BAAI large model family bucket

2024 06

FlagOpen 2.0 large-scale model full-stack open-source technology base

2024 03

Beijing International Dialogue on AI Safety

2023 06

Biological macromolecule structure prediction model OpenComplex

2023 02

FlagOpen 1.0 large model technology open-source system

2022 07

WuDao 3.0 series of models fully open-sourced

2022 06

Jiuding Intelligent Computing Platform

2021 09

Open platform for the large model of WuDao

2021 06

WuDao 2.0 The world's largest pre-trained model at the time

2021 03

WuDao 1.0 China's first ultra-large-scale information intelligence model

2020 10

Set up WuDao Team to start the research and development of large models

2019 04

Hundred of BAAI scholars started free exploration

2018 11

Beijing Academy of Artificial Intelligence was established

News

Press release

Media coverage

智源开源多模态向量模型BGE-VL：多模态检索新突破

在图文检索、组合图像检索等主要多模态检索任务中取得了最佳效果

DeepSeek和OpenAI、xAI、Anthropic哪家强？FlagEval智源评测

2025年伊始，多家顶尖大模型企业如DeepSeek、xAI、OpenAI和Anthropic相继推出了革新性产品。其中，DeepSeek-R1、Grok3、OpenAI的o1、o3和GPT-4.5-preview、Claude-3.7-sonnet等模型引发了业界的广泛关注和热烈讨论。

首次实现DeepSeek-R1满血版多种芯片高效快速部署

构建多元芯片的统一技术栈和开源软硬件生态

智源研究院发布2025十大AI技术趋势

在2025年伊始，智源研究院提出十个人工智能技术及应用趋势，以剖析科技演进轨迹，洞察未来发展方向，预测关键驱动力，指引人类社会迈向更加智能、美好与互联的未来。

智源发布异构统一通信库FlagCX，填补多元算力开源软件栈重要拼图

在多算力时代，通信库作为算力大规模聚合的基础软件，迎来了两个新的重要挑战。

智源研究院与安谋科技达成战略合作，共建开源AI“芯”生态

12月25日，智源研究院与安谋科技（中国）有限公司（以下简称“安谋科技”）正式签署战略合作协议，双方将面向多元AI芯片领域开展算子库优化与适配、编译器与工具链支持、生态系统建设与推广等一系列深入合作，共同打造基于Arm架构的开源技术生态体系，赋能国内大模型与人工智能产业的高速发展。

科技推动力·大模型“智”在何方数字线虫“天宝” 开启生物智能新范式

科技推动力·大模型“智”在何方

市政协委员解读全球首个原生多模态世界模型“Emu3”

市政协委员解读全球首个原生多模态世界模型“Emu3”，北京智源人工智能研究院正式发布全球首个通用人工智能体，诞生了全球首个原生多模态世界模型“Emu3”。

万字实录：生数、智谱、宇树、智源对谈AI的共识与非共识｜甲子光年

2024年，AI的发展方向似乎变得不再明朗。乐观的人依然相信，AI浪潮才刚刚开始，需要给市场与创业者一定的时间与耐心。

共同推动AI治理

科技日报——共同推动AI治理

李开复周志华纵论AI大模型，商汤徐立倡议「打脸时刻」，万字梳理MEET'25大咖激辩，320万观众同见证

「Scaling Law」和「打脸时刻」，绝对是2024年科技智能领域的年度关键词。

Join Us

Openness

Innovation

Excellence

Responsibility

Research

Large Language Model Multimodal Large Model Life Science Large Model Embodied Large Model

System

FlagOpen FlagOS FlagEval FlagData

Ecosystem

BAAI Scholars BAAI Conference BAAI Community AI Safety

Industry

BAAI Innovation Center Partner

media@baai.ac.cn

Contact Number:

© 2022 Beijing Academy of Artificial Intelligence 京ICP备19012194号-1

京公网安备11010802045062