🕒 refactor: Use Legacy Content for Custom Endpoints and Azure Serverless for Improved Compatibility (#8502)

* 🕒 refactor: Use Legacy Content for Custom Endpoints to Improve Compatibility

- Also applies to Azure serverless endpoints from AI Foundry

* chore: move useLegacyContent condition before early return

* fix: Ensure useLegacyContent is set only when options are available

api/server/services/Endpoints/openAI/initialize.js

@@ -65,19 +65,20 @@ const initializeClient = async ({
   const isAzureOpenAI = endpoint === EModelEndpoint.azureOpenAI;
   /** @type {false | TAzureConfig} */
   const azureConfig = isAzureOpenAI && req.app.locals[EModelEndpoint.azureOpenAI];
-
+  let serverless = false;
   if (isAzureOpenAI && azureConfig) {
     const { modelGroupMap, groupMap } = azureConfig;
     const {
       azureOptions,
       baseURL,
       headers = {},
-      serverless,
+      serverless: _serverless,
     } = mapModelToAzureConfig({
       modelName,
       modelGroupMap,
       groupMap,
     });
+    serverless = _serverless;
 
     clientOptions.reverseProxyUrl = baseURL ?? clientOptions.reverseProxyUrl;
     clientOptions.headers = resolveHeaders(
@@ -143,6 +144,9 @@ const initializeClient = async ({
     clientOptions = Object.assign({ modelOptions }, clientOptions);
     clientOptions.modelOptions.user = req.user.id;
     const options = getOpenAIConfig(apiKey, clientOptions);
+    if (options != null && serverless === true) {
+      options.useLegacyContent = true;
+    }
     const streamRate = clientOptions.streamRate;
     if (!streamRate) {
       return options;