Describe the bug

Using Azure models, the cached_content_token_count is logged, when turning on debug logging in litellm. In python-adk the usage_metadata section for caching is not populated.

To Reproduce

• Configure a model: e.g. LiteLlm( model="azure/gpt-4.1" )
• Use the model for an agent with a prompt with at least 2000 tokens two times
• When using the agent the second time the usage_metadata.cached_content_token_count should be filled, but is null at the moment. (I do not want to spam this issue with 2000 tokens....)

Expected behavior
I would like to see the usage_metadata filled with the correct data to make valid cost estimates at runtime.

Screenshots
nothing to screenshot here besides debugger sessions with None objects...

Desktop (please complete the following information):

• OS: macOS
• Python version(python -V): 3.13.7
• ADK version(pip show google-adk): 1.15.1

Model Information:

• Are you using LiteLLM: Yes
• Which model is being used: Azure GPT4.1 mostly

Additional context

The problem is probably at these lines:

• https://github.com/google/adk-python/blob/main/src/google/adk/models/lite_llm.py#L466
• https://github.com/google/adk-python/blob/main/src/google/adk/models/lite_llm.py#L494
• https://github.com/google/adk-python/blob/main/src/google/adk/models/lite_llm.py#L873
  where the cached token usage is not set.