AI Chat
wa_llm
AI WhatsApp bot that follows group chats and replies when mentioned.
★ 171 starsMIT
View guide →HuixiangDou — LLM knowledge assistant built for group chats, with rejection so it answers only when it should
HuixiangDou is an open-source ai chat project by InternLM (InternLM/HuixiangDou on GitHub). It has 2,502 GitHub stars and 179 forks, is released under the BSD-3-Clause license, and was last updated on GitHub on 24 November 2025.
Decide the deployment shape: group-chat assistant (chat_in_group) or streaming repo chat, and your config tier — CPU-only via API, standard 2 GB, or 10 GB multimodal.
Clone the repo and install the system packages for document parsing plus the Python requirements.
Put your documents in repodir, then run the store service to build features into workdir and set the positive/negative example thresholds in config.ini.
Set your model and API key (or a local vLLM server) in config.ini, run the main test to see on-topic answers and off-topic rejections, then launch the Gradio UI or API server.
Preprocess → rejection → response stages are designed for group chat: answer when the knowledge base applies, stay silent otherwise.
Works from your documents plus good/bad question examples that tune the rejection thresholds.
Dense retrieval for documents, sparse for code, plus knowledge-graph and optional internet search and image+text retrieval in the multimodal edition.
Supports DeepSeek, InternLM, GLM, Kimi, StepFun, vLLM, SiliconCloud and others, over Excel/HTML/Markdown/PDF/PPT/TXT/Word documents.
WeChat (Android and wkteam), Lark/Feishu two-way group integration, OpenXLab web app, Gradio demo, HTTP API server and ReadTheDocs embedding.
Documented CPU-only (API-based), standard 2 GB GPU and 10 GB multimodal configurations.
Clone HuixiangDou and enter the directory.
git clone https://github.com/InternLM/HuixiangDou.git
cd HuixiangDouInstall the parsing system packages listed in the README, then the Python requirements.
pip install -r requirements.txtCopy example (or your own) documents into repodir, build the knowledge base, then run the Q&A test after setting your LLM in config.ini.
mkdir repodir && cp -rf resource/data* repodir/
mkdir workdir
python3 -m huixiangdou.services.store
python3 -m huixiangdou.mainInternLM (Shanghai AI Laboratory's open-source team) maintains it; HuixiangDou2 is the newer GraphRAG successor linked from the README.
Yes, BSD-3-Clause licensed. WeChat integration notes in the README mention the integration itself can carry costs in some setups; deploy it yourself for the free path.
Typically yes for the LLM unless you self-host: the standard config uses an OpenAI-style API (Kimi, DeepSeek, StepFun...) or your own local vLLM server, and CPU-only mode uses the SiliconCloud API.
Generic RAG answers every message. HuixiangDou's rejection stage and good/bad question tuning are built specifically so a bot in a group chat only speaks when the question is in scope.
Clone, pip install -r requirements.txt, build the knowledge base from repodir with huixiangdou.services.store, set the LLM in config.ini, and run python3 -m huixiangdou.main.
AI WhatsApp bot that follows group chats and replies when mentioned.
Telegram-summary-bot — summarize group chat with ai, llm && query group chat, open-source.
A smart WhatsApp bot that automates chats and adds AI features.
A feature-rich, self-hostable ChatGPT app built with Next.js.
Open-source AI customer support with live chat and RAG knowledge base.
Open-source Telegram shop bot with crypto payments built in.
The most-starred other AI Chat projects in the LibreAI catalogue — the realistic shortlist if HuixiangDou is not the right fit.
✨ Zero-config AI chat assistant. No API key needed — sign up and instantly chat with GPT-5, Claude 4, Gemini 2.5, DeepSeek & 100+ top models. Pay-as-you-go s…
AI productivity studio with smart chat, autonomous agents and 300+ assistants for frontier LLMs
Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI…
Powerful AI Client
A self-hosted AI answering engine — Perplexity-style search with citations, on your own server
Free and open — grab the source, star the repo, and start creating.