diff --git a/litellm/main.py b/litellm/main.py
index 8d8930ca49..3b3539ee1a 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -1000,9 +1000,11 @@ def completion(
             logging.pre_call(
                 input=prompt, api_key=None, additional_args={"api_base": api_base, "custom_prompt_dict": litellm.custom_prompt_dict}
             )
-            if kwargs.get('acompletion', False) == True:
-                async_generator = ollama.async_get_ollama_response_stream(api_base, model, prompt, optional_params)
-                return async_generator
+            if kwargs.get('acompletion', False) == True:    
+                if optional_params.get("stream", False) == True:
+                # assume all ollama responses are streamed
+                    async_generator = ollama.async_get_ollama_response_stream(api_base, model, prompt, optional_params)
+                    return async_generator
 
             generator = ollama.get_ollama_response_stream(api_base, model, prompt, optional_params)
             if optional_params.get("stream", False) == True: