Overview

Mixen API 1.0

OpenAI-compatible API for Mixen AI. Drop-in: point any OpenAI client at this base URL and use your Mixen key — nothing else changes.

from openai import OpenAI

client = OpenAI(base_url="https://api.mixen.ai/v1", api_key="mxn-...")
resp = client.chat.completions.create(
    model="gpt-5.5",
    messages=[{"role": "user", "content": "Hello!"}],
)

Auth — Authorization: Bearer mxn-.... Create keys in your cabinet.
Billing — from your Mixen balance.
Rate limit — 60 requests/min per key by default (429 + Retry-After).
Endpoints — chat/completions, images/generations, audio/speech (TTS), audio/transcriptions (STT), embeddings, models.
Models — GET /models returns the live catalog with display names, pricing (RUB + USD, identical to what is charged) and capabilities. Pass the id (or a common OpenAI/Claude/Gemini name, e.g. gpt-4o, claude-opus) in the model field.

Wire protocols

The same models and the same key are reachable over four request formats — pick the one your client already speaks:

Protocol	Base URL	Auth header	Used by
OpenAI Chat Completions	`https://api.mixen.ai/v1`	`Authorization: Bearer`	OpenAI SDK, most tools
OpenAI Responses	`https://api.mixen.ai/v1`	`Authorization: Bearer`	Codex CLI, OpenAI Agents SDK
Anthropic Messages	`https://api.mixen.ai`	`Authorization: Bearer`	Claude Code, Anthropic SDK
Google Gemini	`https://api.mixen.ai/v1beta`	`x-goog-api-key`	gemini-cli, google-genai SDK

POST /responses, POST /messages and POST /messages/count_tokens are listed below. The Gemini surface is not part of this schema — its routes are addressed Google-style (POST /v1beta/models/{model}:generateContent, :streamGenerateContent, plus GET /v1beta/models), which OpenAPI paths cannot express; see the gemini-cli guide.

Information

OpenAPI version: 3.1.0

Authentication

ApiKeyAuth

Security scheme type: http

Bearer format: mxn-...

Overview

Mixen API 1.0

Wire protocols

Operations

Authentication

ApiKeyAuth