Yunxin API Documentation

Welcome to Yunxin

Yunxin is a unified AI API gateway. It puts 24 model providers and hundreds of models behind a single endpoint and a single API key — so you write your integration once and switch models with a one-line change. Text, reasoning, vision, embeddings, images, audio, video, music, and 3D all share the same gateway.

Why Yunxin?

• One integration, every model. Use the OpenAI SDK, the Anthropic SDK, or plain HTTP. Change the model parameter to switch providers — no other code changes.
• Three native API formats. OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages are all first-class — each is adapted to its provider faithfully, not bolted on.
• Truly multi-modal. Beyond text: image generation, TTS/STT, video generation, music, and 3D — through consistent, predictable endpoints.
• Built-in reliability. Automatic provider fallback (with transparent X-Fallback-* headers), smart routing, and per-request usage accounting.
• Runtime model registry. Providers and models are configured at runtime and synced from upstream — new models can appear without a redeploy. Always discover what's live with the Models API.

The three request formats

Base URL

All API requests go to:

Authenticate every request with an API key — see Authentication. New to the platform? The Quickstart gets you to a first response in a few minutes.

Discover models at runtime

Yunxin's model catalog is dynamic. Rather than hardcoding model lists, query the live catalog and read each model's capabilities to see what it supports (vision, reasoning, function calling, and more):

curl https://api.yuhuanstudio.com/v1/models \
  -H "Authorization: Bearer $YUNXIN_API_KEY"

See the Models API for filtering by provider, type, and capability.