### Check for existing issues

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

### The Feature

Creating this issue to link to PR that will be submited shortly.

Azure OpenAI Deployment Id is not underlying model name. If we leverage base_model we can tell LiteLLM exactly which model is being used, this fixes an existing issue with Azure OpenAI model detection logic which breaks if deployment names are not standard. This feature proposes a backward compatible solution that unblocks Azure OpenAI customers.

``` import litellm

# =========================================================================== # BEFORE: Custom deployment name — LiteLLM can't detect model type # ===========================================================================

# ❌ This uses default Azure config because "my-gpt5-deployment" doesn't match # any known model pattern. max_tokens won't be mapped to max_completion_tokens, # and gpt-5-specific behavior is lost.

# --- SDK direct call (before) --- response = litellm.completion( model="azure/gpt5_series/my-gpt5-deployment", messages=[{"role": "user", "content": "Hello!"}], max_tokens=100, api_k…