智谱开源可操控手机的智能体模型：AI手机只在一家做是不够的

Core Insights - The company Zhiyu AI, part of the "Big Model Six Tigers," has announced the open-sourcing of its core AI Agent model, AutoGLM, aiming to enable AI to effectively "use mobile phones" [1][5] - Zhiyu emphasizes that the development of AI mobile capabilities should not be limited to a few companies, as this could restrict innovation and user access to essential devices [1][2] Company Overview - Zhiyu AI, founded in 2019 and incubated by a team from Tsinghua University, has gained recognition in the AI sector with its GLM model series [4] - The company has completed over ten rounds of financing, raising more than 10 billion RMB, with significant investments from firms like Hillhouse Capital and major internet companies such as Meituan, Alibaba, Tencent, and Xiaomi [5] Product Development - The AutoGLM model was first introduced in October last year, capable of executing a complete operation chain on real devices, including tasks like sending mobile red envelopes [1][4] - The upcoming AutoGLM 2.0, set for release in 2025, will be integrated into a virtual phone that operates in the cloud, allowing for action replay, auditing, and sensitive data isolation [1] Open Source Strategy - The open-sourcing of AutoGLM includes the core trained model, Phone Use capability framework, tools, and a demo covering over 50 Chinese apps, along with an adaptation layer for Android [2] - This strategy allows enterprises and developers to maintain control over data and privacy while utilizing the technology in a compliant environment [2] Industry Trends - The development of AI mobile technology is seen as an irreversible trend, with industry leaders advocating for an open approach to enhance user experience [5] - Recent collaborations, such as between Doubao and Nubia, highlight the growing interest in AI mobile applications, despite ongoing discussions about their challenges [5]