LLMChat：融合玻璃擬態設計的企業級本地大語言模型聊天平台

LLMChat is an open-source local LLM chat application developed by anomixer (hosted on GitHub). It combines modern glassmorphism design with enterprise-grade local deployment capabilities, ensuring data privacy by storing all user data and conversations locally. Key features include support for over 20 mainstream LLM providers (cloud and local), real-time web search, intelligent voice interaction, and a smooth conversation experience. Built with React 18, TypeScript, Node.js, and Ollama, it's designed for on-premise deployment to meet strict data privacy requirements.

Against the backdrop of increasingly strict data privacy regulations, LLMChat addresses the limitations of cloud-based AI services by prioritizing 'data autonomy'. Unlike cloud services that upload data to third-party servers, all user data and conversation records are stored locally, making it ideal for enterprises needing on-premise AI solutions to protect sensitive information. Its design focuses on balancing aesthetic UI (glassmorphism) with functional completeness and data security.

• Glassmorphism UI: Semi-transparent frosted glass effect, soft shadows, gradient backgrounds, and adaptive themes (light/dark/system) for an immersive visual experience.
• Conversation Experience: Real-time streaming responses, Markdown rendering, one-click code copy, quick model switching, and keyboard shortcuts (e.g., Ctrl/Cmd+I for new chat, Ctrl/Cmd+K to clear).
• Voice Interaction: Multi-language speech-to-text (STT) and text-to-speech (TTS) with a smart playback queue.
• Real-time Search: Supports 5 languages (Traditional Chinese, Simplified Chinese, English, Japanese, Korean) and can query weather, news, exchange rates, and real-time stock information.

LLMChat supports over 20 LLM providers, covering:

• Cloud APIs: OpenAI, Anthropic Claude, Google Gemini, xAI Grok
• Cost-effective/open-source APIs: Groq, Mistral, Moonshot AI (Kimi), Together AI, NVIDIA NIM
• API gateways: OpenRouter, Kilo Gateway, Vercel AI Gateway
• Local/self-hosted solutions: Ollama, vLLM, SGLang, LM Studio Additional features: Adjustable context size, automatic vision model detection, and smart Base64 image transmission.

• Authentication: Email verification for user registration (SMTP configurable). The first registered user becomes an admin automatically (no email verification).
• Data Storage: Conversations are stored in separate files under server/data/, supporting up to 50MB file uploads (text, images, PDFs). Admins can back up/restore data by copying files.
• Admin Controls: User CRUD, password reset, and role management (only admins can add users if SMTP is not configured).

• Tech Stack: Frontend (React18, TypeScript, Vite, Lucide React, React i18next); Backend (Node.js18+, Express, Ollama SDK, Nodemailer).
• Deployment:

1. Dev mode: npm install → npm start (runs at localhost:3000).
2. Docker deployment: Build image → run container with port 8080 exposed, volume mapping for data.

• Hardware Requirements: At least 8GB RAM for local LLM runs; lower requirements for cloud API mode.

LLMChat is suitable for:

1. Enterprise Knowledge Management: On-premise deployment to connect internal knowledge bases.
2. Sensitive Industries: Finance, medical, legal sectors needing strict data privacy.
3. Development Teams: Compare performance of different LLMs.
4. Education & Research: Local deployment reduces costs for academic institutions.

LLMChat represents a trend of combining LLM capabilities with data autonomy. Its glassmorphism design offers a premium user experience, while multi-provider support avoids vendor lock-in. For enterprises evaluating AI deployment, it provides a secure, feature-rich, and ready-to-use solution. As data privacy regulations tighten and demand for autonomous AI grows, local deployment solutions like LLMChat are expected to gain more adoption.