This issue requests support for TensorRT LLM backend. No maintainer has responded, so the first step is to elicit a response. If approved, review existing backend integrations in FastChat (e.g., 'fastchat/serve') and study TensorRT LLM documentation to design the integration.