### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

In the LiteLLM dashboard request log detail view, requests with non-zero provider prompt-cache tokens are currently displayed in a confusing and misleading way.

For example, the redacted request metadata contains:

- usage_object.prompt_tokens = 430798 - usage_object.completion_tokens = 47 - usage_object.prompt_tokens_details.cached_tokens = 430464 - model_map_value.input_cost_per_token = 5.9e-7 - model_map_value.cache_read_input_token_cost = 1.5e-7 - model_map_value.output_cost_per_token = 0.0000020699999999999997

The cost calculation itself appears correct, but the UI hides the cache-hit/cache-miss split and shows Cache Hit: False, even though cached_tokens > 0.

Current display:

```text Tokens: 430,845 (430,798 prompt tokens + 47 completion tokens) Cache Hit: False Input Cost: $0.06476666 (430,798 prompt tokens) Output Cost: $0.00009729 (47 completion tokens) Original LLM Cost: $0.06486395 ```

This makes users think provider prompt caching did not work, or that the request was billed incorrectly.

Expected display:

```text Tokens: 430,8…