In `litellm/proxy/management_endpoints/usage_endpoints/ai_usage_chat.py`, both `stream_usage_ai_chat()` and `_stream_final_response()` call `litellm.acompletion(model=<alias>, ...)` directly. Bare `litellm.acompletion` only infers a provider from a real provider-prefixed model string — it has no concept of virtual proxy aliases registered in `model_list`. Since the Usage dashboard's "Ask AI" model dropdown is populated from exactly those virtual aliases, every real selection fails with:

``` litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=<alias> ```

which the endpoint's broad except-all catches and reports as a generic "An internal error occurred. Please try again." Only the hardcoded `DEFAULT_COMPETITOR_DISCOVERY_MODEL` fallback ever works, and it isn't selectable from the UI.

**Repro:** configure any virtual model alias in `model_list`, open the Usage dashboard, select it in "Ask AI", send a message → generic error; confirm via proxy logs for the `BadRequestError`.

**Suggested fix:** resolve through `llm_router.acompletion()` when the requested model is a registered alias (`llm_router.get_model_list(model_n…