diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
index f7ca2a89048..19464dfd2a0 100644
--- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
+++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
@@ -4725,7 +4725,7 @@ def test_route_image_generation_cost_falls_back_to_requested_size(monkeypatch, r
     assert cost == expected_cost
 
 
-def test_generic_cost_per_token_bills_reasoning_nested_in_text_tokens_once(_local_model_cost_map):
+def test_generic_cost_per_token_bills_reasoning_nested_in_text_tokens_once(_local_model_cost_map: None) -> None:
     """
     Realtime usage (OpenAI and Azure) reports output_tokens == text_tokens + audio_tokens with
     reasoning_tokens already counted inside text_tokens, so reasoning must not be billed on top.
@@ -4757,7 +4757,7 @@ def test_generic_cost_per_token_bills_reasoning_nested_in_text_tokens_once(_loca
     )
 
 
-def test_generic_cost_per_token_keeps_billing_reasoning_reported_beside_text_tokens(_local_model_cost_map):
+def test_generic_cost_per_token_keeps_billing_reasoning_reported_beside_text_tokens(_local_model_cost_map: None) -> None:
     """Providers whose text_tokens exclude reasoning (text + reasoning == completion) stay billed in full."""
 
     model = "gpt-realtime-2.1-mini"
diff --git a/tests/test_litellm/test_cost_calculator.py b/tests/test_litellm/test_cost_calculator.py
index 203f0a9e840..3eb051982e4 100644
--- a/tests/test_litellm/test_cost_calculator.py
+++ b/tests/test_litellm/test_cost_calculator.py
@@ -4494,7 +4494,7 @@ def test_batch_cost_calculator_gpt_6_astra_bills_half_the_standard_rate(_local_m
     assert completion_cost == pytest.approx(500 * 2.5e-5)
 
 
-def test_handle_realtime_stream_cost_calculation_bills_nested_reasoning_tokens_once(_local_model_cost_map):
+def test_handle_realtime_stream_cost_calculation_bills_nested_reasoning_tokens_once(_local_model_cost_map: None) -> None:
     """Realtime response.done nests reasoning_tokens inside text_tokens, so they are billed once."""
     results: OpenAIRealtimeStreamList = [
         {"type": "session.created", "session": {"model": "gpt-realtime-2.1-mini"}},