innightwolfsleep/llm_telegram_bot

Supports multiple LLM backends (llama.cpp, exllama, ollama, transformers) with pluggable generator architecture, enabling flexible model deployment. Provides rich Telegram UX through command prefixes for impersonation/editing, character/model switching mid-conversation, voice generation via Silero, and image generation via Stable Diffusion API. Runs standalone, as a text-generation-webui extension, or on Google Colab with separate per-user session history and multi-mode operation (chat, notebook, query, admin).

127 stars.