diff --git a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py
index f60fd0166..77e723679 100644
--- a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py
+++ b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py
@@ -388,6 +388,7 @@ async def pass_through_request(  # noqa: PLR0915
             _parsed_body=_parsed_body,
             passthrough_logging_payload=passthrough_logging_payload,
             litellm_call_id=litellm_call_id,
+            request=request,
         )
         # done for supporting 'parallel_request_limiter.py' with pass-through endpoints
         logging_obj.update_environment_variables(
@@ -567,6 +568,7 @@ async def pass_through_request(  # noqa: PLR0915
 
 
 def _init_kwargs_for_pass_through_endpoint(
+    request: Request,
     user_api_key_dict: UserAPIKeyAuth,
     passthrough_logging_payload: PassthroughStandardLoggingPayload,
     _parsed_body: Optional[dict] = None,
@@ -582,6 +584,12 @@ def _init_kwargs_for_pass_through_endpoint(
     }
     if _litellm_metadata:
         _metadata.update(_litellm_metadata)
+
+    _metadata = _update_metadata_with_tags_in_header(
+        request=request,
+        metadata=_metadata,
+    )
+
     kwargs = {
         "litellm_params": {
             "metadata": _metadata,
@@ -593,6 +601,13 @@ def _init_kwargs_for_pass_through_endpoint(
     return kwargs
 
 
+def _update_metadata_with_tags_in_header(request: Request, metadata: dict) -> dict:
+    _tags = request.headers.get("tags")
+    if _tags:
+        metadata["tags"] = _tags.split(",")
+    return metadata
+
+
 def create_pass_through_route(
     endpoint,
     target: str,
diff --git a/tests/pass_through_tests/test_local_vertex.js b/tests/pass_through_tests/test_local_vertex.js
index 7ae9b942a..9ee603e7a 100644
--- a/tests/pass_through_tests/test_local_vertex.js
+++ b/tests/pass_through_tests/test_local_vertex.js
@@ -1,31 +1,22 @@
 const { VertexAI, RequestOptions } = require('@google-cloud/vertexai');
 
 
-// Import fetch if the SDK uses it
-const originalFetch = global.fetch || require('node-fetch');
-
-// Monkey-patch the fetch used internally
-global.fetch = async function patchedFetch(url, options) {
-    // Modify the URL to use HTTP instead of HTTPS
-    if (url.startsWith('https://localhost:4000')) {
-        url = url.replace('https://', 'http://');
-    }
-    console.log('Patched fetch sending request to:', url);
-    return originalFetch(url, options);
-};
 
 const vertexAI = new VertexAI({
     project: 'adroit-crow-413218',
     location: 'us-central1',
-    apiEndpoint: "localhost:4000/vertex-ai"
+    apiEndpoint: "127.0.0.1:4000/vertex-ai"
 });
 
+// Create customHeaders using Headers
+const customHeaders = new Headers({
+    "X-Litellm-Api-Key": "sk-1234",
+    tags: "vertexjs,test-2"
+});
 
 // Use customHeaders in RequestOptions
 const requestOptions = {
-    customHeaders: new Headers({
-        "x-litellm-api-key": "sk-1234"
-    })
+    customHeaders: customHeaders,
 };
 
 const generativeModel = vertexAI.getGenerativeModel(
@@ -33,7 +24,7 @@ const generativeModel = vertexAI.getGenerativeModel(
     requestOptions
 );
 
-async function streamingResponse() {
+async function testModel() {
     try {
         const request = {
             contents: [{role: 'user', parts: [{text: 'How are you doing today tell me your name?'}]}],
@@ -49,20 +40,4 @@ async function streamingResponse() {
     }
 }
 
-
-async function nonStreamingResponse() {
-    try {
-        const request = {
-            contents: [{role: 'user', parts: [{text: 'How are you doing today tell me your name?'}]}],
-          };
-        const response = await generativeModel.generateContent(request);
-        console.log('non streaming response: ', JSON.stringify(response));
-    } catch (error) {
-        console.error('Error:', error);
-    }
-}
-
-
-
-streamingResponse();
-nonStreamingResponse();
\ No newline at end of file
+testModel();
\ No newline at end of file