Talk to your LiteLLM proxy

Set URL and an sk-… key in the rail, pick a model, then send a prompt. Replies show here.