Overview

Base URL, authentication, endpoints, and errors for the SERV Reasoning API.

The SERV Reasoning API is a REST API at https://inference-api.openserv.ai. It implements the OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages formats, so existing clients work with a base URL and key change. To get running, start with the Quickstart.

Authentication

Generate an API key at console.openserv.ai and send it as a bearer token:

Authorization: Bearer $SERV_API_KEY

Endpoints

Method and path	Format	Reference
POST /v1/chat/completions	OpenAI	Chat completions
POST /v1/responses	OpenAI	Responses
POST /v1/messages	Anthropic	Messages

/v1/chat/completions works with every model in the catalog. /v1/responses is OpenAI models only. /v1/messages accepts most of the catalog — see endpoint compatibility.

System prompt required

Every request must include a system prompt. Where it goes depends on the endpoint — a system message for chat completions, top-level instructions for responses, or top-level system for messages. Requests without one are rejected.

Errors

The API uses standard HTTP status codes. Error bodies follow the format of the endpoint’s upstream API.

Status	Meaning
400	Invalid request — missing system prompt, unsupported model, or malformed body.
401	Missing or invalid API key.
404	Unknown path or model.
429	Rate limited.
5xx	Server or upstream provider error.

{
  "error": {
    "type": "invalid_request_error",
    "message": "A system prompt is required. Please include a system or developer message in your request."
  }
}