feat: updated current tests

tests/models/test_litellm_chatcompletions_stream.py

@@ -55,7 +55,7 @@ async def test_stream_response_yields_events_for_text_content(monkeypatch) -> No
             prompt_tokens=7,
             total_tokens=12,
             completion_tokens_details=CompletionTokensDetails(reasoning_tokens=2),
-            prompt_tokens_details=PromptTokensDetails(cached_tokens=5),
+            prompt_tokens_details=PromptTokensDetails(cached_tokens=6),
         ),
     )
 
@@ -122,6 +122,8 @@ async def patched_fetch_response(self, *args, **kwargs):
     assert completed_resp.usage.input_tokens == 7
     assert completed_resp.usage.output_tokens == 5
     assert completed_resp.usage.total_tokens == 12
+    assert completed_resp.usage.input_tokens_details.cached_tokens == 6
+    assert completed_resp.usage.output_tokens_details.reasoning_tokens == 2
 
 
 @pytest.mark.allow_call_model_methods

tests/test_openai_chatcompletions.py

@@ -13,7 +13,10 @@
     ChatCompletionMessageToolCall,
     Function,
 )
-from openai.types.completion_usage import CompletionUsage
+from openai.types.completion_usage import (
+    CompletionUsage,
+    PromptTokensDetails,
+)
 from openai.types.responses import (
     Response,
     ResponseFunctionToolCall,
@@ -51,7 +54,13 @@ async def test_get_response_with_text_message(monkeypatch) -> None:
         model="fake",
         object="chat.completion",
         choices=[choice],
-        usage=CompletionUsage(completion_tokens=5, prompt_tokens=7, total_tokens=12),
+        usage=CompletionUsage(
+            completion_tokens=5,
+            prompt_tokens=7,
+            total_tokens=12,
+            # completion_tokens_details left blank to test default
+            prompt_tokens_details=PromptTokensDetails(cached_tokens=3),
+        ),
     )
 
     async def patched_fetch_response(self, *args, **kwargs):
@@ -81,6 +90,8 @@ async def patched_fetch_response(self, *args, **kwargs):
     assert resp.usage.input_tokens == 7
     assert resp.usage.output_tokens == 5
     assert resp.usage.total_tokens == 12
+    assert resp.usage.input_tokens_details.cached_tokens == 3
+    assert resp.usage.output_tokens_details.reasoning_tokens == 0
     assert resp.response_id is None
 
 
@@ -127,6 +138,8 @@ async def patched_fetch_response(self, *args, **kwargs):
     assert resp.usage.requests == 0
     assert resp.usage.input_tokens == 0
     assert resp.usage.output_tokens == 0
+    assert resp.usage.input_tokens_details.cached_tokens == 0
+    assert resp.usage.output_tokens_details.reasoning_tokens == 0
 
 
 @pytest.mark.allow_call_model_methods

tests/test_openai_chatcompletions_stream.py

@@ -8,7 +8,11 @@
     ChoiceDeltaToolCall,
     ChoiceDeltaToolCallFunction,
 )
-from openai.types.completion_usage import CompletionUsage
+from openai.types.completion_usage import (
+    CompletionTokensDetails,
+    CompletionUsage,
+    PromptTokensDetails,
+)
 from openai.types.responses import (
     Response,
     ResponseFunctionToolCall,
@@ -46,7 +50,13 @@ async def test_stream_response_yields_events_for_text_content(monkeypatch) -> No
         model="fake",
         object="chat.completion.chunk",
         choices=[Choice(index=0, delta=ChoiceDelta(content="llo"))],
-        usage=CompletionUsage(completion_tokens=5, prompt_tokens=7, total_tokens=12),
+        usage=CompletionUsage(
+            completion_tokens=5,
+            prompt_tokens=7,
+            total_tokens=12,
+            prompt_tokens_details=PromptTokensDetails(cached_tokens=2),
+            completion_tokens_details=CompletionTokensDetails(reasoning_tokens=3),
+        ),
     )
 
     async def fake_stream() -> AsyncIterator[ChatCompletionChunk]:
@@ -112,6 +122,8 @@ async def patched_fetch_response(self, *args, **kwargs):
     assert completed_resp.usage.input_tokens == 7
     assert completed_resp.usage.output_tokens == 5
     assert completed_resp.usage.total_tokens == 12
+    assert completed_resp.usage.input_tokens_details.cached_tokens == 2
+    assert completed_resp.usage.output_tokens_details.reasoning_tokens == 3
 
 
 @pytest.mark.allow_call_model_methods