fix: 三端联调测试 2 P0 + 6 P1 + 2 P2 修复

P0-1: SaaS relay 模型别名解析 — "glm-4-flash" → "glm-4-flash-250414" (resolve_model)
P0-2: config.rs interpolate_env_vars UTF-8 修复 (chars 迭代器替代 bytes as char)
      + DB 启动编码检查 + docker-compose UTF-8 编码参数

P1-3: UI 模型选择器覆盖 Agent 默认模型 (model_override 全链路: TS→Tauri→Rust kernel)
P1-6: 知识搜索管道修复 — seed_knowledge 创建 chunks + 默认分类 (seed/uploaded/distillation)
P1-7: 用量限额从当前 Plan 读取 (非 stale usage 表)
P1-8: relay 双维度配额检查 (relay_requests + input_tokens)

P2-9: SSE 路径 token 计数修复 — 流结束检测替代固定 500ms sleep + billing increment

desktop/src-tauri/src/kernel_commands/chat.rs

@@ -30,6 +30,9 @@ pub struct ChatRequest {
     /// Enable sub-agent delegation (Ultra mode only)
     #[serde(default)]
     pub subagent_enabled: Option<bool>,
+    /// Model override — UI 选择的模型优先于 Agent 配置的默认模型
+    #[serde(default)]
+    pub model: Option<String>,
 }
 
 /// Chat response
@@ -76,6 +79,9 @@ pub struct StreamChatRequest {
     /// Enable sub-agent delegation (Ultra mode only)
     #[serde(default)]
     pub subagent_enabled: Option<bool>,
+    /// Model override — UI 选择的模型优先于 Agent 配置的默认模型
+    #[serde(default)]
+    pub model: Option<String>,
 }
 
 // ---------------------------------------------------------------------------
@@ -116,7 +122,7 @@ pub async fn agent_chat(
         None
     };
 
-    let response = kernel.send_message_with_chat_mode(&id, request.message, chat_mode)
+    let response = kernel.send_message_with_chat_mode(&id, request.message, chat_mode, request.model)
         .await
         .map_err(|e| format!("Chat failed: {}", e))?;
 
@@ -257,6 +263,7 @@ pub async fn agent_chat_stream(
             prompt_arg,
             session_id_parsed,
             Some(chat_mode_config),
+            request.model.clone(),
         )
             .await
             .map_err(|e| {

desktop/src/lib/kernel-chat.ts

@@ -60,6 +60,7 @@ export function installChatMethods(ClientClass: { prototype: KernelClient }): vo
       reasoning_effort?: string;
       plan_mode?: boolean;
       subagent_enabled?: boolean;
+      model?: string;
     }
   ): Promise<{ runId: string }> {
     const runId = crypto.randomUUID();
@@ -199,6 +200,7 @@ export function installChatMethods(ClientClass: { prototype: KernelClient }): vo
           reasoningEffort: opts?.reasoning_effort,
           planMode: opts?.plan_mode,
           subagentEnabled: opts?.subagent_enabled,
+          model: opts?.model,
         },
       });
     } catch (err: unknown) {

desktop/src/store/chat/streamStore.ts

@@ -303,6 +303,9 @@ export const useStreamStore = create<StreamState>()(
         .map(m => ({ role: m.role, content: m.content }))
         .slice(-20);
 
+      // UI 模型选择器应覆盖 Agent 默认模型
+      const currentModel = useConversationStore.getState().currentModel;
+
       const result = await client.chatStream(
         content,
         {
@@ -534,6 +537,7 @@ export const useStreamStore = create<StreamState>()(
             set({ isStreaming: false, activeRunId: null });
           },
         },
+        // eslint-disable-next-line @typescript-eslint/no-explicit-any
         {
           sessionKey: effectiveSessionKey,
           agentId: effectiveAgentId,
@@ -542,7 +546,8 @@ export const useStreamStore = create<StreamState>()(
           plan_mode: get().getChatModeConfig().plan_mode,
           subagent_enabled: get().getChatModeConfig().subagent_enabled,
           history,
-        }
+          model: currentModel || undefined,
+        } as Parameters<typeof client.chatStream>[2]
       );
 
       if (result?.runId) {