证券日报网讯 12月17日，岩山科技在互动平台回答投资者提问时表示，公司旗下控股子公司岩芯数智 于2025年7月在世界人工智能大会上公开发布了Yan2.0Preview模型，Yan2.0Preview新增了原生"记忆模 块"，在不依赖外挂知识库的情况下具备了参数化的记忆能力，将推动下游应用的各种设备向自主学习 迈进；同时将多模态能力进一步拓展到视频领域，可实时理解学习短视频中的行为动作、物体状态等信 息，推动大模型与现实环境的深度交互。未来，RockAI将重点围绕"Yan架构"持续加强核心算法创新及 迭代升级，以"离线智能"为支点，着力构建"Yan架构"智能生态。 （文章来源：证券日报） ...

Core Insights - In the first half of 2025, the company reported a revenue of 317 million yuan, a year-on-year decrease of 6.38%, while the net profit attributable to shareholders increased by 81.51% to 67.11 million yuan, and the net profit after deducting non-recurring items rose by 159.06% to 58.08 million yuan [1] Group 1: Business Developments - The company aims to build a smarter, better, and more harmonious world centered around artificial intelligence, aspiring to become a global leader in AI and technological innovation [1][3] - Nullmax, a key player in the smart driving sector, has developed a comprehensive auxiliary driving application and established deep partnerships with several well-known automotive manufacturers for mass production projects [1][2] - The company has achieved TISAX AL3 certification, ensuring its information security management meets the highest standards in the European automotive industry, facilitating global market expansion [2] Group 2: Research and Innovation - The Shanghai Yansi Brain Research Institute has made significant advancements in brain-computer interface and brain-like intelligence, focusing on decoding algorithms and non-organic brain disease diagnostics [2] - A milestone achievement in brain-computer interface technology was reached with the development of a four-level decoding architecture, enabling direct interpretation of neural signals into complete sentences [2] - In the generative AI (AIGC) sector, the company launched the Yan2.0Preview model, which includes a memory module and extends multimodal capabilities to video, enhancing interaction between large models and real-world environments [2][3] Group 3: Commercialization Efforts - RockAI has begun providing localized applications and deployments for enterprise clients, addressing needs for data privacy, security, and cost-effective solutions [3] - Collaborations with various terminal device manufacturers, including robots and PCs, are being explored to implement the Yan model across multiple downstream devices [3] - A generative smart medical record system based on the Yan architecture has been successfully launched in hospitals, improving the quality and efficiency of medical record writing [3]