Workflow
科学大模型
icon
Search documents
学会“主动思考” 紫东太初4.0来了 28家企业签约成为生态合作伙伴
Chang Jiang Ri Bao· 2025-09-20 00:46
Core Insights - The launch of the ZDTC 4.0 multimodal reasoning model and the "ZDTC Cloud" platform aims to promote the industrialization of domestic large models, with 28 companies signing agreements to become ecosystem partners [2][5]. Group 1: Model Development - The ZDTC 4.0 model represents a significant milestone in the evolution of multimodal large models, transitioning from passive analysis to active reasoning capabilities [3]. - The model can integrate multimodal information and generate explainable intelligent decisions, showcasing its ability to respond in seconds to user queries, such as providing travel information based on a photo [3][4]. - Compared to the previous version, ZDTC 3.0, the 4.0 version demonstrates enhanced fine-grained multimodal interaction and reasoning abilities, allowing it to mimic human attention mechanisms [3][4]. Group 2: Applications and Impact - The ZDTC model has been applied in over 100 scenarios, including a smart service platform for orthopedic instruments that manages thousands of medical supplies [5]. - The "ZDTC Cloud" platform offers comprehensive support for enterprises, covering computing power services, model training, application development, and embodied intelligence [5]. - The model has been benchmarked against international standards in multimodal understanding and logical reasoning, achieving breakthroughs in tasks such as long video comprehension [4]. Group 3: Future Prospects - The company aims to create more benchmark cases in low-altitude economy, embodied intelligence, manufacturing, and healthcare, with a revenue target of 100 million yuan by 2025 [5]. - The establishment of partnerships with various organizations, including the "AI + Meteorology" innovation studio, indicates a strategic focus on collaborative development in AI technology [7].
人民播客——“人工智能+”行动解读① 科研正从“大海捞针”走向“精准导航”?
Ren Min Wang· 2025-09-18 06:00
这释放了什么信号?AI到底怎样颠覆传统的科研模式?"科学大模型"和我们熟悉的ChatGPT、DeepSeek 有啥不一样?AI怎么打破数学、物理、化学等"学科壁垒"?未来5到10年,科研形式会发生翻天覆地的 变化吗? 本期嘉宾来自AI for Science(人工智能赋能科学技术)领域的先行者——北京科学智能研究院,我们邀 请到了研究院院长李鑫宇,他将以生动的语言,带我们踏上一次"科研未来之旅",深入了解这场正在发 生的"科研范式革命"。 本期嘉宾:北京科学智能研究院院长李鑫宇 近期国务院发布的《关于深入实施人工智能+行动的意见》(以下简称《意见》),提出加快实施六大 重点行动,"人工智能+科学技术"排在首位。 对话AI摘编: 主持人:国务院《关于深入实施"人工智能+"行动的意见》将"人工智能+科学技术"放在六大行动首 位,在您看来这传递了什么信号? 李鑫宇:首先国家敏锐把握到了它对社会发展的重要意义。从历史角度看,"科学技术是第一生产力"一 直是核心,把它放在首位和整体发展逻辑一致,它是"AI+各行业、各领域"的底层支撑。这是重新梳理 二者关系的重要政策——这次是把科学技术重新拉回视野,找回"以科技突破为底 ...