About wenda
Wenda (闻达) is an open-source LLM calling platform designed for efficient content generation in specific environments. It addresses the computational resource limitations of individuals and small-to-medium enterprises, while ensuring knowledge security and privacy. The platform integrates a knowledge base supporting local offline vector libraries, local search engines, and online search engines. It supports a wide range of large language models including chatGLM-6B/chatGLM2-6B, chatRWKV, Llama, MOSS, Baichuan, Aquila-7B, InternLM, and online APIs such as OpenAI API and chatGLM-130b API. Additional features include Auto scripts (JavaScript plugins for custom workflows), conversation history management, intranet deployment, multi-user concurrent access, streaming output, quantization, and LoRA support.
Key Features
Pros & Cons
- Free and open-source
- Supports a wide range of LLMs (both offline and online APIs)
- Integrated knowledge base for retrieval-augmented generation
- Extensible with JavaScript Auto scripts
- Privacy-focused: can be deployed on intranet, no data leakage
- Works on low-resource hardware (e.g., 6GB VRAM)
- Documentation primarily in Chinese, may be less accessible for non-Chinese speakers
- Some models (Llama, MOSS, Baichuan without LoRA) are not recommended for Chinese users
- Requires manual setup and configuration (Python environment, model downloads)
- Limited community support outside Chinese ecosystem