llmcord turns Discord into an LLM frontend supporting virtually any language model, whether remotely hosted via OpenAI API, xAI, Google Gemini, Mistral, Groq, or OpenRouter, or locally via Ollama, LM Studio, or vLLM. It provides a reply-chain conversation system where users @mention the bot to start threads, branch discussions, and chain consecutive messages. The bot supports vision models for image attachments and includes features like automatic conversation continuation in DMs and thread branching.