Kirara AI

Kirara AI is an open-source, do-it-yourself multimodal AI chatbot framework that brings mainstream large language models into the chat platforms people already use, including QQ, Telegram, Discord, WeChat and Feishu. Originally started in 2022, it has grown into a full bot platform with a web admin console, visual workflows and a plugin marketplace, and it has earned about 19,100 stars on GitHub.

Overview

Kirara AI connects one bot to many model backends, including DeepSeek, Grok, Claude, Gemini, OpenAI models, Qwen, Kimi and local models via Ollama, and to drawing models such as Stable Diffusion, Flux and Midjourney for AI image generation. A graphical WebUI handles model management, workflow design and plugin installation, while persona presets let you tune the bot’s character, from everyday assistant to role-play companion. It supports multi-turn conversation, keyword and conditional triggers, admin commands, voice replies through text-to-speech engines, and an HTTP API so other systems can talk to the bot programmatically. Deployment is self-hosted, with Docker support and built-in tunneling helpers.

Key Features

• One bot across QQ, Telegram, Discord, WeChat, WeCom, Feishu and OneBot platforms
• Many LLM backends: DeepSeek, Claude, Gemini, OpenAI, Groq-style APIs and local Ollama models
• Multimodal chat with AI drawing (Stable Diffusion, Flux, Midjourney) and voice replies
• Visual workflow system, persona presets and a plugin marketplace for extensions
• Web admin console for model management, triggers, commands and multi-account support
• HTTP API for integrating the bot with external applications and services

💳 Pricing

Kirara AI is free and open source under the AGPL-3.0 license. You host it yourself and only pay for any third-party model APIs you connect.

Who Is It For

Kirara AI is for community managers, hobbyists and developers who want a customizable AI companion or assistant living inside their group chats and messaging apps.