UnioAPI Documentation

Model API Overview

Understand model identifiers, routing and protocol compatibility in UnioAPI. Find API quickstart guides, coding tool setup, SDK examples and endpoint references.

UnioAPI is a unified model API service for AI teams: call many models through one API entry point, one API key, and the OpenAI-compatible or Anthropic-compatible protocols you already know. Protocol adaptation, routing, and failover are handled by the platform.

Three concepts

Model identifiers

The model in your request is a Unio model identifier (for example gpt-5.6-sol), and the same identifier is echoed back in responses. How it maps to upstream models is not exposed. See GET /v1/models for the list of callable models.

Routes and candidates

A single model may have multiple supply channels (candidates) inside the platform. The platform picks among them based on real-time health; when the current candidate cannot complete a request, it tries the next available candidate for the same model — the requested model never changes.

Protocol compatibility

The platform serves each protocol in its native shape and does not translate one public protocol into another:

Protocol familyEndpointsNotes
OpenAI-compatiblePOST /v1/chat/completions, POST /v1/responses, GET /v1/modelsOpenAI request, response, and streaming shapes
Anthropic-compatiblePOST /v1/messagesAnthropic request, response, and named SSE events

Brand names only indicate the compatible protocol family and do not imply official partnership or endorsement.

Start here

  • 5-minute quickstart: get a key, change one base_url, make your first call.
  • Client setup: step-by-step for Claude Code, Codex, Cursor, TRAE, and OpenCode.
  • API reference: endpoints, parameters, streaming, and error semantics.
  • Platform guides: billing basics, common invoice questions, and rate limits.

On this page