diff --git a/llm_server/routes/v1/generate_stream.py b/llm_server/routes/v1/generate_stream.py
index 2e2c188..7e97acc 100644
--- a/llm_server/routes/v1/generate_stream.py
+++ b/llm_server/routes/v1/generate_stream.py
@@ -104,10 +104,12 @@ def stream(ws):
 
                     end_time = time.time()
                     elapsed_time = end_time - start_time
+                    r_headers = dict(request.headers)
+                    r_url = request.url
 
                     def background_task():
                         generated_tokens = tokenize(generated_text)
-                        log_prompt(handler.client_ip, handler.token, input_prompt, generated_text, elapsed_time, handler.parameters, dict(request.headers), response_status_code, request.url, response_tokens=generated_tokens)
+                        log_prompt(handler.client_ip, handler.token, input_prompt, generated_text, elapsed_time, handler.parameters, r_headers, response_status_code, r_url, response_tokens=generated_tokens)
 
                     # TODO: use async/await instead of threads
                     threading.Thread(target=background_task).start()
diff --git a/server.py b/server.py
index 27885c1..fab5727 100644
--- a/server.py
+++ b/server.py
@@ -20,7 +20,6 @@ from llm_server.routes.server_error import handle_server_error
 # TODO: add more excluding to SYSTEM__ tokens
 # TODO: make sure the OpenAI moderation endpoint scans the last n messages rather than only the last one (make that threaded)
 # TODO: support turbo-instruct on openai endpoint
-# TODO: show requested model (not actual LLM backend model) in OpenAI responses
 
 try:
     import vllm