🚀 OpenAPI-manager

One API gateway for global AI providers, regional relay, load balancing, Edge Proxy acceleration, and multimodal traffic.

OpenAPI-manager gives apps and agents one OpenAI-compatible endpoint while the server handles provider keys, routing, retries, RPM/TPM pressure, regional forwarding, and provider compatibility.

🌍 Languages

• 🇺🇸 English
• 🇨🇳 简体中文
• 🇹🇼 繁體中文
• 🇯🇵 日本語
• 🇰🇷 한국어
• 🇸🇦 العربية
• 🇹🇭 ไทย
• 🇪🇸 Español
• 🇧🇷 Português do Brasil
• 🇫🇷 Français
• 🇩🇪 Deutsch
• 🇮🇩 Bahasa Indonesia
• 🇻🇳 Tiếng Việt

✨ Quick Picture

Apps / Agents / Users
        ↓
OpenAPI-manager
        ↓
OpenAI / Claude / Gemini / Groq / OpenRouter / NVIDIA
DeepSeek / Qwen / ERNIE / Hunyuan / Doubao / GLM / Kimi / SiliconFlow / ...

🐳 Docker Hub

heing8848/openapi-manager

docker pull heing8848/openapi-manager:latest
docker run --name openapi-manager \
  -d \
  --restart always \
  -p 3000:3000 \
  -e TZ=Asia/Shanghai \
  -v /home/ubuntu/data/openapi-manager:/data \
  heing8848/openapi-manager:latest

Open http://localhost:3000.

Default first login: root / 123456. Change it immediately.

🧭 Core Ideas

• 🌐 Users can be in China or overseas; provider requests leave from your OpenAPI-manager server region.
• ⚖️ Multi-key and multi-channel load balancing helps with strict RPM/TPM limits.
• ⚡ Optional Cloudflare Edge Proxy can speed up latency-sensitive streaming routes.
• 🧩 One OpenAI-compatible API surface can front many global and China-friendly providers.
• 🖼️ Multimodal relay paths cover chat, vision, image, audio, and video-task style providers where supported.

🙏 Attribution

OpenAPI-manager is based on one-api by songquanpeng. Thanks to the original author and contributors for the foundation.

📄 License

MIT License. See LICENSE.