diff --git a/litellm/main.py b/litellm/main.py index f30d93361e5..3c06f5adfc4 100644 --- a/litellm/main.py +++ b/litellm/main.py @@ -6054,7 +6054,7 @@ def completion( custom_llm_provider=custom_llm_provider, model_id=optional_params.get("model_id"), ) - if mock_response or mock_tool_calls or mock_timeout: + if mock_response is not None or mock_tool_calls or mock_timeout: kwargs.pop("mock_timeout", None) # remove for any fallbacks triggered return mock_completion( model, diff --git a/tests/unit/test_utils.py b/tests/unit/test_utils.py index 55b7bc78a11..1a4f0eafbeb 100644 --- a/tests/unit/test_utils.py +++ b/tests/unit/test_utils.py @@ -157,6 +157,19 @@ def test_post_call_rule_rejects_mock_response(monkeypatch: pytest.MonkeyPatch) - ) +def test_completion_empty_string_mock_response_does_not_make_request() -> None: + response = litellm.completion( + model="openai/gpt-4o-mini", + messages=[{"role": "user", "content": "hi"}], + mock_response="", + api_base="http://127.0.0.1:9", + api_key="sk-test", + num_retries=0, + ) + + assert response.choices[0].message.content == "" + + def test_post_call_rule_rejects_streaming_mock_response(monkeypatch: pytest.MonkeyPatch) -> None: monkeypatch.setattr(litellm, "pre_call_rules", []) monkeypatch.setattr(litellm, "post_call_rules", [_reject_short_post_call_response])