diff --git a/ai-service-admin/src/api/llm.ts b/ai-service-admin/src/api/llm.ts
new file mode 100644
index 0000000..73b04f5
--- /dev/null
+++ b/ai-service-admin/src/api/llm.ts
@@ -0,0 +1,50 @@
+import request from '@/utils/request'
+import type {
+  LLMProviderInfo,
+  LLMConfig,
+  LLMConfigUpdate,
+  LLMTestResult,
+  LLMTestRequest,
+  LLMProvidersResponse,
+  LLMConfigUpdateResponse
+} from '@/types/llm'
+
+export function getLLMProviders(): Promise<LLMProvidersResponse> {
+  return request({
+    url: '/llm/providers',
+    method: 'get'
+  })
+}
+
+export function getLLMConfig(): Promise<LLMConfig> {
+  return request({
+    url: '/llm/config',
+    method: 'get'
+  })
+}
+
+export function saveLLMConfig(data: LLMConfigUpdate): Promise<LLMConfigUpdateResponse> {
+  return request({
+    url: '/llm/config',
+    method: 'put',
+    data
+  })
+}
+
+export function testLLM(data: LLMTestRequest): Promise<LLMTestResult> {
+  return request({
+    url: '/llm/test',
+    method: 'post',
+    data
+  })
+}
+
+export type {
+  LLMProviderInfo,
+  LLMConfig,
+  LLMConfigUpdate,
+  LLMTestResult,
+  LLMTestRequest,
+  LLMProvidersResponse,
+  LLMConfigUpdateResponse
+}
diff --git a/ai-service-admin/src/api/rag.ts b/ai-service-admin/src/api/rag.ts
index 1e0147b..cd6c495 100644
--- a/ai-service-admin/src/api/rag.ts
+++ b/ai-service-admin/src/api/rag.ts
@@ -1,9 +1,131 @@
 import request from '@/utils/request'
 
-export function runRagExperiment(data: { query: string, kbIds?: string[], params?: any }) {
+export interface AIResponse {
+  content: string
+  prompt_tokens?: number
+  completion_tokens?: number
+  total_tokens?: number
+  latency_ms?: number
+  model?: string
+}
+
+export interface RetrievalResult {
+  content: string
+  score: number
+  source: string
+  metadata?: Record<string, any>
+}
+
+export interface RagExperimentRequest {
+  query: string
+  kb_ids?: string[]
+  top_k?: number
+  score_threshold?: number
+  llm_provider?: string
+  generate_response?: boolean
+}
+
+export interface RagExperimentResult {
+  query: string
+  retrieval_results?: RetrievalResult[]
+  final_prompt?: string
+  ai_response?: AIResponse
+  total_latency_ms?: number
+}
+
+export function runRagExperiment(data: RagExperimentRequest): Promise<RagExperimentResult> {
   return request({
     url: '/admin/rag/experiments/run',
     method: 'post',
     data
   })
 }
+
+export function runRagExperimentStream(
+  data: RagExperimentRequest,
+  onMessage: (event: MessageEvent) => void,
+  onError?: (error: Event) => void,
+  onComplete?: () => void
+): EventSource {
+  const baseUrl = import.meta.env.VITE_APP_BASE_API || '/api'
+  const url = `${baseUrl}/admin/rag/experiments/stream`
+  
+  const eventSource = new EventSource(url, {
+    withCredentials: true
+  })
+  
+  eventSource.onmessage = onMessage
+  eventSource.onerror = (error) => {
+    eventSource.close()
+    onError?.(error)
+  }
+  
+  return eventSource
+}
+
+export function createSSEConnection(
+  url: string,
+  body: RagExperimentRequest,
+  onMessage: (data: string) => void,
+  onError?: (error: Error) => void,
+  onComplete?: () => void
+): () => void {
+  const baseUrl = import.meta.env.VITE_APP_BASE_API || '/api'
+  const fullUrl = `${baseUrl}${url}`
+  
+  const controller = new AbortController()
+  
+  fetch(fullUrl, {
+    method: 'POST',
+    headers: {
+      'Content-Type': 'application/json',
+      'Accept': 'text/event-stream',
+    },
+    body: JSON.stringify(body),
+    signal: controller.signal
+  })
+    .then(async (response) => {
+      if (!response.ok) {
+        throw new Error(`HTTP error! status: ${response.status}`)
+      }
+      
+      const reader = response.body?.getReader()
+      if (!reader) {
+        throw new Error('No response body')
+      }
+      
+      const decoder = new TextDecoder()
+      let buffer = ''
+      
+      while (true) {
+        const { done, value } = await reader.read()
+        
+        if (done) {
+          onComplete?.()
+          break
+        }
+        
+        buffer += decoder.decode(value, { stream: true })
+        const lines = buffer.split('\n')
+        buffer = lines.pop() || ''
+        
+        for (const line of lines) {
+          if (line.startsWith('data: ')) {
+            const data = line.slice(6)
+            if (data === '[DONE]') {
+              onComplete?.()
+              return
+            }
+            onMessage(data)
+          }
+        }
+      }
+    })
+    .catch((error) => {
+      if (error.name !== 'AbortError') {
+        onError?.(error)
+      }
+    })
+  
+  return () => controller.abort()
+}
diff --git a/ai-service-admin/src/types/llm.ts b/ai-service-admin/src/types/llm.ts
new file mode 100644
index 0000000..50889df
--- /dev/null
+++ b/ai-service-admin/src/types/llm.ts
@@ -0,0 +1,42 @@
+export interface LLMProviderInfo {
+  name: string
+  display_name: string
+  description?: string
+  config_schema: Record<string, any>
+}
+
+export interface LLMConfig {
+  provider: string
+  config: Record<string, any>
+  updated_at?: string
+}
+
+export interface LLMConfigUpdate {
+  provider: string
+  config?: Record<string, any>
+}
+
+export interface LLMTestResult {
+  success: boolean
+  response?: string
+  latency_ms?: number
+  prompt_tokens?: number
+  completion_tokens?: number
+  total_tokens?: number
+  message?: string
+  error?: string
+}
+
+export interface LLMTestRequest {
+  test_prompt?: string
+  config?: LLMConfigUpdate
+}
+
+export interface LLMProvidersResponse {
+  providers: LLMProviderInfo[]
+}
+
+export interface LLMConfigUpdateResponse {
+  success: boolean
+  message: string
+}
diff --git a/docs/progress/ai-service-admin-progress.md b/docs/progress/ai-service-admin-progress.md
index 43b074d..5b220d0 100644
--- a/docs/progress/ai-service-admin-progress.md
+++ b/docs/progress/ai-service-admin-progress.md
@@ -50,7 +50,7 @@ version: "0.3.0"
 - [x] (P5-01~P5-06) 后端管理接口实现
 
 #### Phase 6: 嵌入模型管理（待处理）
-- [ ] (P5-01) API 服务层与类型定义 [AC-ASA-08, AC-ASA-09]
+- [x] (P5-01) API 服务层与类型定义 [AC-ASA-08, AC-ASA-09]
 - [ ] (P5-02) 提供者选择组件 [AC-ASA-09]
 - [ ] (P5-03) 动态配置表单 [AC-ASA-09, AC-ASA-10]
 - [ ] (P5-04) 测试连接组件 [AC-ASA-11, AC-ASA-12]
@@ -60,7 +60,7 @@ version: "0.3.0"
 - [ ] (P5-08) 组件整合与测试 [AC-ASA-08~AC-ASA-13]
 
 #### Phase 7: LLM 配置与 RAG 调试输出（当前）
-- [ ] (P6-01) LLM API 服务层与类型定义：创建 src/api/llm.ts 和 src/types/llm.ts [AC-ASA-14, AC-ASA-15]
+- [x] (P6-01) LLM API 服务层与类型定义：创建 src/api/llm.ts 和 src/types/llm.ts [AC-ASA-14, AC-ASA-15]
 - [ ] (P6-02) LLM 提供者选择组件：创建 LLMProviderSelect.vue [AC-ASA-15]
 - [ ] (P6-03) LLM 动态配置表单：创建 LLMConfigForm.vue [AC-ASA-15, AC-ASA-16]
 - [ ] (P6-04) LLM 测试连接组件：创建 LLMTestPanel.vue [AC-ASA-17, AC-ASA-18]
diff --git a/spec/ai-service-admin/tasks.md b/spec/ai-service-admin/tasks.md
index d274e43..f7791cb 100644
--- a/spec/ai-service-admin/tasks.md
+++ b/spec/ai-service-admin/tasks.md
@@ -123,7 +123,7 @@ principles:
 
 > 页面导向：嵌入模型配置页面，支持提供者切换、参数配置、连接测试。
 
-- [ ] (P5-01) API 服务层与类型定义：创建 src/api/embedding.ts 和 src/types/embedding.ts
+- [x] (P5-01) API 服务层与类型定义：创建 src/api/embedding.ts 和 src/types/embedding.ts
   - AC: [AC-ASA-08, AC-ASA-09]
 
 - [ ] (P5-02) 提供者选择组件：实现 `EmbeddingProviderSelect` 下拉组件，对接 `/admin/embedding/providers`
@@ -153,7 +153,7 @@ principles:
 
 | 任务 | 描述 | 状态 |
 |------|------|------|
-| P5-01 | API 服务层与类型定义 | ⏳ 待处理 |
+| P5-01 | API 服务层与类型定义 | ✅ 已完成 |
 | P5-02 | 提供者选择组件 | ⏳ 待处理 |
 | P5-03 | 动态配置表单 | ⏳ 待处理 |
 | P5-04 | 测试连接组件 | ⏳ 待处理 |
@@ -170,7 +170,7 @@ principles:
 
 ### 6.1 LLM 模型配置
 
-- [ ] (P6-01) LLM API 服务层与类型定义：创建 src/api/llm.ts 和 src/types/llm.ts
+- [x] (P6-01) LLM API 服务层与类型定义：创建 src/api/llm.ts 和 src/types/llm.ts
   - AC: [AC-ASA-14, AC-ASA-15]
 
 - [ ] (P6-02) LLM 提供者选择组件：实现 `LLMProviderSelect` 下拉组件
@@ -208,7 +208,7 @@ principles:
 
 | 任务 | 描述 | 状态 |
 |------|------|------|
-| P6-01 | LLM API 服务层与类型定义 | ⏳ 待处理 |
+| P6-01 | LLM API 服务层与类型定义 | ✅ 已完成 |
 | P6-02 | LLM 提供者选择组件 | ⏳ 待处理 |
 | P6-03 | LLM 动态配置表单 | ⏳ 待处理 |
 | P6-04 | LLM 测试连接组件 | ⏳ 待处理 |