diff --git a/litellm/proxy/cost.log b/litellm/proxy/cost.log
new file mode 100644
index 0000000000..566d7dbbb1
--- /dev/null
+++ b/litellm/proxy/cost.log
@@ -0,0 +1,4 @@
+2023-10-09 14:46:28 - Model gpt-3.5-turbo-0613 Cost: 6.1e-05
+2023-10-09 14:46:29 - Model gpt-3.5-turbo Cost: 0.0
+2023-10-09 14:48:18 - Model gpt-3.5-turbo-0613 Cost: 0.00004700
+2023-10-09 14:48:18 - Model gpt-3.5-turbo Cost: 0.00000000
diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py
index e0a8ae68fe..ca1ee6f237 100644
--- a/litellm/proxy/proxy_server.py
+++ b/litellm/proxy/proxy_server.py
@@ -22,6 +22,7 @@ from fastapi import FastAPI, Request
 from fastapi.routing import APIRouter
 from fastapi.responses import StreamingResponse, FileResponse
 import json
+import logging
 
 app = FastAPI()
 router = APIRouter()
@@ -205,11 +206,32 @@ async def chat_completion(request: Request):
         final_prompt_value=os.getenv("MODEL_POST_PROMPT", "")
     )
     response = litellm.completion(**data)
+
+    # track cost of this response, using litellm.completion_cost
+    await track_cost(response)
     if 'stream' in data and data['stream'] == True: # use generate_responses to stream responses
         return StreamingResponse(data_generator(response), media_type='text/event-stream')
     print_verbose(f"response: {response}")
     return response
 
+async def track_cost(response):
+    try:
+        logging.basicConfig(
+            filename='cost.log',
+            level=logging.INFO,
+            format='%(asctime)s - %(message)s',
+            datefmt='%Y-%m-%d %H:%M:%S'
+        )
+        import datetime
+
+        response_cost = litellm.completion_cost(completion_response=response)
+
+        logging.info(f"Model {response.model} Cost: {response_cost:.8f}")
+
+    except:
+        pass
+
+
 
 @router.get("/ollama_logs")
 async def retrieve_server_log(request: Request):