Docs

Five minutes from signup to first token.

Kunavo follows the OpenAI wire protocol. If you're already using the OpenAI SDK, your migration cost is one base_url change. Really.

Start the quickstart Browse 35 models

Your first call

Point the SDK at Kunavo.

The OpenAI Python SDK works as-is. Set base_url and an API key from /app/keys — your existing chat / image / embedding code keeps working with any frontier model.

first_call.py

from openai import OpenAI

client = OpenAI(
    api_key="sk-kn-...",
    base_url="https://api.kunavo.com/v1",
)

resp = client.chat.completions.create(
    model="claude-haiku-4-5",
    messages=[{"role": "user", "content": "Hello, Kunavo"}],
)
print(resp.choices[0].message.content)

Reference

Pick a topic

Getting started

Quickstart

Integrations

Setup for 53 clients — Claude Code, Cline, Kilo Code, Roo Code, Codex CLI, opencode, CC Switch, SillyTavern, Janitor AI and more: base URL, key and model name.

Authentication

API keys, base URL, header conventions, and key rotation best practices.

Claude API key

Get a Claude API key, call Claude with the OpenAI SDK or the native Messages API, and the error codes a bad key returns.

Gemini API key

Get a Gemini API key from Google AI Studio, call it with the OpenAI SDK, and keep it out of your bundle.

Billing

How wallet top-up, the markup formula, and the credit ledger work.

Usage API

GET /v1/usage — pull daily token counts and cost for any API key, scoped to that key only.

Errors

HTTP status codes, error payload shape, and a retry strategy that won't burn budget.

Endpoints

Chat completions

OpenAI-compatible /v1/chat/completions with streaming, tools, vision and reasoning.

Messages API

Native Anthropic Messages API — pass cache_control, tools and thinking straight through to Claude.

Images

Text-to-image and image-edit (i2i) — Nano Banana and GPT-Image-2.

Video

Veo 3.1 — Fast, Quality and Lite. Long-running task API.

Music

Suno V5 — text-to-music with vocals, lyrics and styles. Async task API.

Guides

Prompt caching

Cut Anthropic input cost by 90% on repeat prompts — how cache_control, affinity routing, and dashboard metrics all wire together.

Ollama OpenAI API

Call Ollama's local OpenAI-compatible API — and point the same SDK at hosted Claude and GPT.

Migrating from OpenAI

soon

Drop-in migration guides from OpenAI, Anthropic, Google direct.