LS
灵枢大模型LingShu Model
负责认知、推理与生成。Responsible for cognition, reasoning, and generation.
PRD / C1引核 CortexCortex
AIOS 的智能如何统一运行？How does AIOS intelligence run as one system?
引核是 AIOS 的智能系统内核。Cortex is the intelligence kernel of AIOS.
引核统一大模型、算力、推理调度、运行时上下文与 Agent Runtime，让任务理解、执行计划和能力调用可以在同一个智能运行层中发生。Cortex unifies models, compute, reasoning orchestration, runtime context, and Agent Runtime so task understanding, execution planning, and capability invocation happen in one intelligence layer.
引核由三层组成Cortex has three layers
03 项03 SIGNALSTS
天枢算力平台TianShu Compute
负责模型接入、路由和推理服务。Responsible for model access, routing, and inference service.
RT
Cortex RuntimeCortex Runtime
负责智能体运行、上下文、任务执行和能力调用。Responsible for agent runtime, context, task execution, and capability invocation.
01Intelligence RuntimeIntelligence Runtime
引核解决模型、算力、上下文和执行状态各自孤立的问题。Cortex resolves isolated models, compute, context, and execution state.
它不是单一大模型，也不是模型网关；它是 AIOS 运行智能任务时需要的统一智能调度与运行时机制。It is not a single large model or a model gateway; it is the unified intelligence orchestration and runtime mechanism AIOS needs for intelligent tasks.
MODEL
大模型能力Model Capability
提供自然语言理解、知识推理、内容生成和智能决策能力。Provides language understanding, knowledge reasoning, generation, and decision support.
ROUTE
模型与算力路由Model and Compute Routing
按任务、成本、性能和场景约束统一接入与调度模型和算力。Routes models and compute by task, cost, performance, and scenario constraints.
STATE
运行时上下文Runtime Context
管理单次任务与持续运行过程中的上下文、工作记忆与执行状态。Manages context, working memory, and execution state during tasks and continuous operation.
02BoundaryBoundary
引核负责执行智能任务，不负责所有系统关系。Cortex runs intelligent tasks; it does not own every system relationship.
长期成长属于元域，网络级匹配与价值分配属于织界，设备身份属于引芯，空间上下文属于引元。Long-term growth belongs to Realm, network matching and value distribution to Orcha, device identity to Nexus, and spatial context to Origin.
REALM
与元域With Realm
引核管理当前运行；元域保存跨任务成长、经验和能力版本。Cortex manages current execution; Realm preserves growth, experience, and capability versions across tasks.
ORCHA
与织界With Orcha
引核执行已确定的调用计划；织界负责网络级匹配、计量和分配。Cortex executes determined invocation plans; Orcha owns network matching, measurement, and distribution.
FIELD
与引芯 / 引元With Nexus / Origin
引芯提供可信设备能力入口，引元提供真实空间上下文，引核提供智能理解与运行时支撑。Nexus provides trusted device capability entry, Origin provides real-space context, and Cortex provides intelligence and runtime support.
Next ProductNext Product
理解智能内核之后，看设备能力如何可信入图。After the intelligence kernel, see how device capabilities enter the graph with trust.
引芯让真实设备具备身份、位置与能力证明。Nexus gives real devices identity, location, and capability proof.