← Frameworks

Spring AISpring AI

Spring（VMware/Broadcom）Spring (VMware/Broadcom)

开源Open source JavaJava SpringBootSpring Boot MCPMCP 向量库Vector store partial lifecycle: active 核验Checked 2026-11-03

Spring AI 把 AI 能力做成 Spring 组件：可移植 Model API、ChatClient、向量库与 MCP 集成，用 Spring Boot 的 starter 和 auto-configuration 装配。

跨大类对照：vLLM。

AI as Spring components: a portable Model API, ChatClient, vector stores and MCP integration, assembled with Spring Boot starters and auto-configuration.

适合与不适合When to use it, when not to

适合：Spring Boot 团队、要把 AI 做成标准 Bean、需要 MCP 与向量库集成。不适合：非 Java 团队、要成品 agent、要轻量脚本调用。

Fits Spring Boot teams who want AI as standard Spring components and need MCP plus vector-store integration. Does not fit non-Java teams, turnkey agents, or lightweight scripting.

中文生态China availability 国产 / 国内可直接用 / 备案状态 · 快照domestic / direct access / CAC filing · snapshot 2026-11-03

出品属地Origin: 国外出品foreign-made　Spring 项目（VMware/Broadcom 生态）。Spring project (VMware/Broadcom ecosystem).
国内使用Direct access: 大陆可直连direct OK　docs.spring.io 大陆直连未核验；Maven Central 可经镜像获取。Direct reachability from mainland not verified; Maven Central is available through mirrors.
备案状态CAC filing: 未见于备案名单not in filing list：未见于网信办公示名单（2026-11-03 检索）；开源框架未面向大陆公众提供服务，备案不适用。Not in the CAC public registry (searched 2026-11-03); the open-source framework does not serve the mainland public, so filing does not apply.

备案信息以网信办《生成式人工智能服务已备案信息》公示名单为准（按批次滚动更新，本站只抄编号不解读）；「大陆可直连」是本站从大陆网络直连实测的结果，不是厂商承诺。GitHub 类站点当日实测有波动，口径写在依据里。Filing status is quoted from the CAC public registry (updated batch by batch; this site copies the number and does not interpret it). "Direct access" is this site's own reachability measurement from a mainland network, not a vendor promise. GitHub-hosted sites fluctuated on the measurement day — the note says so.

固定坐标系Fixed coordinate system

8 个维度dimensions, 与同分区其他对象逐项可比each comparable item by item against others in the same section.

模型与开放条件Model & openness: 可移植的 Model API：用统一接口接多家模型供应商；官方文档列出 ChatClient 与模型 API。当前支持列表以官方文档为准，本站未逐条清点。
A portable Model API behind one interface across providers; official docs list ChatClient and the model APIs. The current supported list follows official docs and is not enumerated here.
运行位置Where it runs: 运行在 Spring Boot 应用内：官方提供 starters 与 auto-configuration，模型与向量库以 Spring Bean 方式注入。
Runs inside a Spring Boot application: official starters and auto-configuration inject models and vector stores as Spring beans.
本地文件Local files: 文档摄取（ETL）与向量库由应用配置；框架本身不托管数据，索引落盘位置与检索链路由所选向量库决定。
Document ingestion (ETL) and vector stores are configured by the application; the framework hosts no data itself.
关机后的任务Tasks after shutdown: 库形态，不内建任务队列；异步与长任务由 Spring 侧自行编排。
A library with no built-in task queue; async and long-running work is orchestrated on the Spring side.
工具与扩展Tools & extensions: 支持工具调用与 MCP：官方文档写明可消费 MCP server，也可把 Spring 服务暴露给 AI 生态。
Supports tool calling and MCP: official docs state it can consume MCP servers and expose Spring services to the AI ecosystem.
上下文与记忆Context & memory: 对话记忆由 ChatClient 与 memory 配置管理；持久化需自己接存储。
Conversation memory is managed through ChatClient and memory configuration; persistence is yours to attach.
权限与限制Permissions & limits: Spring Security 体系内运行；权限由应用配置，无内置多租户隔离。
Runs within Spring Security; permissions are application-configured, with no built-in multi-tenant isolation.
适合什么任务Best for: 适合：Spring Boot 团队、要把 AI 能力做成标准 Spring 组件、需要 MCP 与向量库集成。不适合：非 Java 团队、要成品 agent、要轻量脚本式调用。
Fits Spring Boot teams who want AI as standard Spring components and need MCP plus vector-store integration. Does not fit non-Java teams, turnkey agents, or lightweight scripting.

头号误解Top misconception

把 Spring AI 当成 LangChain4j 的官方版——两者是不同项目，抽象与生态不同Treating Spring AI as the official version of LangChain4j — they are different projects with different abstractions
以为加了 starter 就自动配好模型——仍需 API key 与模型配置Assuming a starter auto-configures a model — you still need an API key and model configuration
忽略 Spring 版本兼容——starters 与 Boot 版本要对齐Ignoring Spring version compatibility — starters and Boot versions must align

未知项清单Known unknowns

当前支持的模型与向量库完整清单（未逐条清点）
MCP 双向集成的版本与限制
Spring Boot 版本兼容矩阵

Full current list of supported models and vector stores (not enumerated here)
Versions and limits of bidirectional MCP integration
Spring Boot compatibility matrix

证据来源Evidence sources

判断可回到以下一手源复核。本站核验日Every judgement can be rechecked against the primary sources below. Verified on 2026-11-03，内容更新日content updated on 2026-11-03。

类型Type	名称Name	链接Link
docs	Spring AI · 官方参考文档（Model API / ChatClient / MCP / ETL；2026-11-03 复核官方页）Spring AI · official reference docs (Model API / ChatClient / MCP / ETL)	https://docs.spring.io/spring-ai/reference/
official	Spring AI · 项目主页（Spring AI 2.0 定位）Spring AI · project home (Spring AI 2.0 positioning)	https://spring.io/projects/spring-ai/
repo	Spring AI · GitHub 仓库Spring AI · GitHub repository	https://github.com/spring-projects/spring-ai
changelog	Spring AI · Releases（官方 changelog）Spring AI · Releases (official changelog)	https://github.com/spring-projects/spring-ai/releases

实测记录Measured runs

本档未跑集成与性能实测；结论只覆盖 API 与官方文档披露的能力面。

This site has not completed integration or performance tests; the conclusions cover the API and capabilities disclosed in official docs only.