celedog 0.1.0

Celedog AI gateway client — OpenAI-compatible, one API for 400+ models

celedog (Python)

OpenAI-compatible Python client for Celedog.io — one API, 400+ AI models, pay as you go.

pip install celedog

Quick start

import celedog

# Reads CELEDOG_API_KEY from the env; or pass it directly:
#   celedog.Client(api_key="sk-celedog-...")
client = celedog.Client()

resp = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello, Celedog!"}],
)
print(resp.choices[0].message.content)

Async + streaming

import asyncio
import celedog

async def main():
    client = celedog.AsyncClient()
    stream = await client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": "Stream me a sentence."}],
        stream=True,
    )
    async for chunk in stream:
        print(chunk.choices[0].delta.content or "", end="", flush=True)

asyncio.run(main())

celedog.Client is a thin subclass of openai.OpenAI — every feature OpenAI ships (multimodal input, structured outputs, tool calling, vision, batch) works unchanged. Just point the SDK at Celedog with your sk-celedog-... key.

Configuration

Source	Variable	Default
Constructor	api_key=	—
Env	CELEDOG_API_KEY	—
Env (fallback)	OPENAI_API_KEY	—
Constructor	base_url=	https://celedog.io/v1
Env	CELEDOG_BASE_URL	https://celedog.io/v1

Other languages

Language	Install
Python	this package
Node.js / TypeScript	npm install celedog
Go	go get github.com/AIRES-Technology/celedog-go
Java	Maven coordinate io.celedog:celedog:0.1.0 (GitHub Packages — see celedog-java)

License

MIT — free to use and integrate. The Celedog gateway server itself is AGPL-3.0 (see celedog.io); SDK code is permissive so you can ship it inside your closed-source app.