Position: It is Time to Virtualize Foundation Models with a Self-evolving Operating System Layer

观点：是时候通过自演进操作系统层来实现基础模型虚拟化了

Abstract: 摘要：

AI applications have shifted from single, monolithic foundation models (FM) to compound agentic systems. Yet today’s stacks remain fragmented: even as protocols (e.g., MCP, A2A) ease tool/agent connectivity, each framework embeds an implicit runtime for state, memory, budgets, and guardrails, making behavior non-portable and governance brittle. 人工智能应用已从单一的、庞大的基础模型（FM）转向复合型智能体系统。然而，当今的技术栈依然处于碎片化状态：尽管各种协议（如 MCP、A2A）简化了工具与智能体之间的连接，但每个框架都嵌入了针对状态、内存、预算和护栏的隐式运行时，这导致了行为的可移植性差以及治理的脆弱性。

It mirrors computing before operating systems, when every program re-implemented basic services. This position paper argues that the field now needs a Foundation Model Operating System (FMOS) — a system layer that virtualizes FM interactions analogous to how virtual machines abstract physical hardware, giving applications the illusion of dedicated, trustworthy FM instances with effectively unbounded capabilities. 这种情况类似于操作系统出现之前的计算时代，当时每个程序都需要重新实现基础服务。本篇观点论文认为，该领域目前需要一个“基础模型操作系统”（FMOS）——即一个能够虚拟化 FM 交互的系统层，其原理类似于虚拟机对物理硬件的抽象，从而为应用程序提供一种错觉，使其仿佛拥有专用且可信的 FM 实例，并具备近乎无限的能力。

Internally, the FMOS orchestrates knowledge across memory tiers, model selection and resource allocation, and verification and policy enforcement. Like the human brain switching between fast intuition and slow deliberation, the FMOS learns when to intervene and when to let inference proceed directly and continuously adapting its policies based on operational experience. 在内部，FMOS 负责协调跨内存层级的知识、模型选择与资源分配，以及验证和策略执行。就像人类大脑在快速直觉与缓慢深思之间切换一样，FMOS 能够学习何时进行干预，何时让推理直接进行，并根据运行经验持续调整其策略。