diff --git a/litellm/tests/test_completion.py b/litellm/tests/test_completion.py
index 07ba97ace..b4f3866b8 100644
--- a/litellm/tests/test_completion.py
+++ b/litellm/tests/test_completion.py
@@ -164,8 +164,8 @@ def test_completion_replicate_stability_stream():
     try:
         response = completion(model=model_name, messages=messages, stream=True, replicate=True)
         # Add any assertions here to check the response
-        for result in response:
-            print(result)
+        for chunk in response:
+            print(chunk['choices'][0]['delta'])
         print(response)
     except Exception as e:
         pytest.fail(f"Error occurred: {e}")
diff --git a/litellm/utils.py b/litellm/utils.py
index b81e9bc0d..b961fe812 100644
--- a/litellm/utils.py
+++ b/litellm/utils.py
@@ -625,13 +625,16 @@ class CustomStreamWrapper:
         return self
 
     def __next__(self):
+        completion_obj ={ "role": "assistant", "content": ""}
         if self.model in litellm.anthropic_models:
           chunk = next(self.completion_stream)
-          return {"choices": [{"delta": chunk.completion}]}
+          completion_obj["content"] = chunk.completion
         elif self.model == "replicate":
            chunk = next(self.completion_stream)
-           return {"choices": [{"delta": chunk}]}
+           completion_obj["content"] = chunk
         elif self.model in litellm.cohere_models:
           chunk = next(self.completion_stream)
-          return {"choices": [{"delta": chunk.text}]}
+          completion_obj["content"] = chunk.text
+        # return this for all models
+        return {"choices": [{"delta": completion_obj}]}