CoAI.Dev

CoAI.Dev is an open-source, Apache 2.0-licensed LLM assets management platform built by CoAI Software Inc. It combines a full-featured AI chat interface for end users with a powerful API gateway and channel management system for developers and enterprises. The project is available on GitHub under the handle coaidev/coai and reached v4.0.0 in October 2025.

What It Is

CoAI.Dev positions itself as an all-in-one AIGC (AI-Generated Content) commercial solution that merges the strengths of frontend-focused chat platforms (like ChatGPT Next Web) and API distribution systems (like One API) into a single deployable stack. It serves three audiences simultaneously: individual users who want a polished AI assistant, developers who need multi-tenant token distribution and API proxying, and enterprises that require elastic cluster deployment, SSO, and compliance tooling. The core platform is written in Go and Rust for the backend and React for the frontend, with MySQL and Redis as data stores.

Architecture and Tech Stack

The platform is structured in three layers:

• AI Application Layer: A built-in chat app, plus API compatibility with tools like Cursor, ComfyUI, Dify, and Coze.
• AI Middleware Layer: An API gateway, auth service, billing system, load balancer, model manager, rate limiter, and analytics engine.
• AI Infrastructure Layer: Connections to third-party model providers (OpenAI, Anthropic, Google Gemini, AWS Bedrock, etc.) and self-hosted models via Ollama, vLLM, LocalAI, and LlamaCpp.

The GitHub README describes the idle memory footprint as approximately 10–15 MB, with a single Docker container deployable via Docker Compose, Kubernetes, Zeabur, BTPanel, or Alibaba Cloud ComputeNest.

Key Capabilities

• Unified API: A single OpenAI-compatible endpoint (/v1/chat/completions, /v1/images, /v1/models) that routes to 35+ providers and 200+ models.
• Intelligent channel management: Priority-based routing, weighted load balancing, automatic failover, model redirection, and channel health monitoring.
• Billing system: Supports both subscription and elastic (pay-per-token or pay-per-request) billing, with Stripe integration, gift codes, and redemption codes.
• File parsing: Out-of-the-box support for PDF, Docx, Pptx, Excel, and image formats; OCR support; S3/R2/MinIO storage backends.
• Web search: Integrated SearXNG engine supporting Google, Bing, DuckDuckGo, Yahoo, Wikipedia, Arxiv, and Qwant.
• Model caching: Request-hash-based caching that skips billing on cache hits, with configurable TTL and result count.
• Conversation sync: Zero-dependency cross-device sync without WebDAV or WebRTC; conversation sharing via link or image.
• Internationalization: UI available in Simplified Chinese, English, Japanese, and Russian.
• PWA + Desktop: Progressive Web App support and a Tauri-based desktop wrapper.

Deployment Model

CoAI.Dev is self-hosted by default. The GitHub README documents four primary deployment paths: Docker Compose (recommended), standalone Docker, Zeabur one-click, and source compilation. The project also supports Kubernetes for distributed deployments and lists Alibaba Cloud ComputeNest as a managed option. A hosted demo is available at app.coai.dev. The commercial and enterprise editions add features such as TTS/STT, a plugin marketplace, RAG knowledge base, SMS/OAuth login, multi-tenant API key distribution, security auditing, and bot integrations (Discord, Telegram, Feishu).

Update: v4.0.0

The latest GitHub release is v4.0.0, published on 2025-10-23. The repository was last pushed in March 2026 and last updated in June 2026, indicating active maintenance. The project reports over 9,200 GitHub stars and 1,200+ forks as of the data collected, and the homepage states the project was "#1 Trending on GitHub" (vendor-published claim). The open-source edition is licensed under Apache 2.0, making it commercially friendly for secondary development and distribution.