Add Gemini/OpenAI token stats to the conversation trace (#141118)

* Add gemini token status to the conversation trace

* Add OpenAI Token Stats

* Revert input_tokens_details since its not in the openai version yet

* Fix ruff lint errors

homeassistant/components/conversation/chat_log.py

@@ -8,7 +8,7 @@ from contextlib import contextmanager
 from contextvars import ContextVar
 from dataclasses import asdict, dataclass, field, replace
 import logging
-from typing import Literal, TypedDict
+from typing import Any, Literal, TypedDict
 
 import voluptuous as vol
 
@@ -456,10 +456,16 @@ class ChatLog:
         LOGGER.debug("Prompt: %s", self.content)
         LOGGER.debug("Tools: %s", self.llm_api.tools if self.llm_api else None)
 
-        trace.async_conversation_trace_append(
-            trace.ConversationTraceEventType.AGENT_DETAIL,
+        self.async_trace(
             {
                 "messages": self.content,
                 "tools": self.llm_api.tools if self.llm_api else None,
-            },
+            }
+        )
+
+    def async_trace(self, agent_details: dict[str, Any]) -> None:
+        """Append agent specific details to the conversation trace."""
+        trace.async_conversation_trace_append(
+            trace.ConversationTraceEventType.AGENT_DETAIL,
+            agent_details,
         )

homeassistant/components/google_generative_ai_conversation/conversation.py

@@ -403,6 +403,18 @@ class GoogleGenerativeAIConversationEntity(
                 error = f"Sorry, I had a problem talking to Google Generative AI: {err}"
                 raise HomeAssistantError(error) from err
 
+            if (usage_metadata := chat_response.usage_metadata) is not None:
+                chat_log.async_trace(
+                    {
+                        "stats": {
+                            "input_tokens": usage_metadata.prompt_token_count,
+                            "cached_input_tokens": usage_metadata.cached_content_token_count
+                            or 0,
+                            "output_tokens": usage_metadata.candidates_token_count,
+                        }
+                    }
+                )
+
             response_parts = chat_response.candidates[0].content.parts
             if not response_parts:
                 raise HomeAssistantError(

homeassistant/components/openai_conversation/conversation.py

@@ -9,6 +9,7 @@ from openai._streaming import AsyncStream
 from openai.types.responses import (
     EasyInputMessageParam,
     FunctionToolParam,
+    ResponseCompletedEvent,
     ResponseFunctionCallArgumentsDeltaEvent,
     ResponseFunctionCallArgumentsDoneEvent,
     ResponseFunctionToolCall,
@@ -111,6 +112,7 @@ def _convert_content_to_param(
 
 
 async def _transform_stream(
+    chat_log: conversation.ChatLog,
     result: AsyncStream[ResponseStreamEvent],
 ) -> AsyncGenerator[conversation.AssistantContentDeltaDict]:
     """Transform an OpenAI delta stream into HA format."""
@@ -137,6 +139,18 @@ async def _transform_stream(
                     )
                 ]
             }
+        elif (
+            isinstance(event, ResponseCompletedEvent)
+            and (usage := event.response.usage) is not None
+        ):
+            chat_log.async_trace(
+                {
+                    "stats": {
+                        "input_tokens": usage.input_tokens,
+                        "output_tokens": usage.output_tokens,
+                    }
+                }
+            )
 
 
 class OpenAIConversationEntity(
@@ -252,7 +266,7 @@ class OpenAIConversationEntity(
                 raise HomeAssistantError("Error talking to OpenAI") from err
 
             async for content in chat_log.async_add_delta_content_stream(
-                user_input.agent_id, _transform_stream(result)
+                user_input.agent_id, _transform_stream(chat_log, result)
             ):
                 messages.extend(_convert_content_to_param(content))
 

tests/components/google_generative_ai_conversation/test_conversation.py

@@ -156,8 +156,10 @@ async def test_function_call(
     trace_events = last_trace.get("events", [])
     assert [event["event_type"] for event in trace_events] == [
         trace.ConversationTraceEventType.ASYNC_PROCESS,
-        trace.ConversationTraceEventType.AGENT_DETAIL,
+        trace.ConversationTraceEventType.AGENT_DETAIL,  # prompt and tools
+        trace.ConversationTraceEventType.AGENT_DETAIL,  # stats for response
         trace.ConversationTraceEventType.TOOL_CALL,
+        trace.ConversationTraceEventType.AGENT_DETAIL,  # stats for response
     ]
     # AGENT_DETAIL event contains the raw prompt passed to the model
     detail_event = trace_events[1]
@@ -166,6 +168,13 @@ async def test_function_call(
         p["tool_name"] for p in detail_event["data"]["messages"][2]["tool_calls"]
     ] == ["test_tool"]
 
+    detail_event = trace_events[2]
+    assert set(detail_event["data"]["stats"].keys()) == {
+        "input_tokens",
+        "cached_input_tokens",
+        "output_tokens",
+    }
+
 
 @patch(
     "homeassistant.components.google_generative_ai_conversation.conversation.llm.AssistAPI._async_get_tools"