chat-demo: Chat directly with the DeepSeek API in the browser, no backend required
deepseek-ai/chat-demo
About the project
A static chat demo leveraging the DeepSeek API that requires no separate application backend server. The browser connects directly to the configured API service. You can try it online immediately, or download the archive from the release, extract it, and double-click index.html to use it locally without any installation or build process.
It manages multiple chat sessions with auto-generated titles and supports prompt editing, answer regeneration, and switching between message branches. Advanced features include streaming responses, deep thinking, smart search, link reading, and source citation. It also provides file attachments, voice input, and voice synthesis, with real-time audio spectrum visualization during recording on mobile.
Supports Markdown, syntax highlighting, KaTeX math formulas, and Mermaid diagram rendering. Compatible with Korean and English, light/dark themes, and mobile layouts. You can back up or migrate data by exporting and importing chat history in ZIP format. API keys and settings are stored in the browser's localStorage, while chat content and attachments are stored in IndexedDB, isolated per browser profile.
By default, it uses the deepseek-flash model via the Anthropic-compatible Messages API. Since it sends requests including extended features such as search, link reading, and title generation, an API service that supports these features is required. Distributed under the MIT license, it can be deployed on a static web server or run directly from the file system, making it suitable for simple prototyping or personal chat clients.
deepseek-ai/chat-demo
Chat Demo
TypeScript
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.