fix bug (trim eats space or \n mistakenly), optimize timeout by model

app/client/platforms/xai.ts

@@ -1,6 +1,6 @@
 "use client";
 // azure and openai, using same models. so using same LLMApi.
-import { ApiPath, XAI_BASE_URL, XAI, REQUEST_TIMEOUT_MS } from "@/app/constant";
+import { ApiPath, XAI_BASE_URL, XAI } from "@/app/constant";
 import {
   useAccessStore,
   useAppConfig,
@@ -17,6 +17,7 @@ import {
   SpeechOptions,
 } from "../api";
 import { getClientConfig } from "@/app/config/client";
+import { getTimeoutMSByModel } from "@/app/utils";
 import { preProcessImageContent } from "@/app/utils/chat";
 import { RequestPayload } from "./openai";
 import { fetch } from "@/app/utils/stream";
@@ -103,7 +104,7 @@ export class XAIApi implements LLMApi {
       // make a fetch request
       const requestTimeoutId = setTimeout(
         () => controller.abort(),
-        REQUEST_TIMEOUT_MS,
+        getTimeoutMSByModel(options.config.model),
       );
 
       if (shouldStream) {