You can install ollama with a script fetched with curl and run a llm model with a grand total of two bash commands (including curl).