Skip to content

Quick Start

Five minutes from install to first response.

1. Set your API key

export OPENAI_API_KEY="sk-..."

# .env
OPENAI_API_KEY=sk-...

from dotenv import load_dotenv
load_dotenv()

resp = completion("gpt-4o-mini", messages, api_key="sk-...")

2. Make your first call

from llmgate import completion

resp = completion(
    "gpt-4o-mini",
    [{"role": "user", "content": "What is the capital of France?"}],
)

print(resp.text)
# → "The capital of France is Paris."

3. Switch providers — literally one word

from llmgate import completion

messages = [{"role": "user", "content": "Explain recursion in one sentence."}]

# OpenAI
resp = completion("gpt-4o-mini", messages)

# Anthropic
resp = completion("claude-3-5-haiku-20241022", messages)

# Gemini
resp = completion("gemini-2.5-flash-lite", messages)

# Groq (fastest)
resp = completion("groq/llama-3.1-8b-instant", messages)

print(resp.text)   # always the same
print(resp.provider)   # "openai" | "anthropic" | "gemini" | "groq"

4. Async

import asyncio
from llmgate import acompletion

async def main():
    resp = await acompletion(
        "groq/llama-3.3-70b-versatile",
        [{"role": "user", "content": "Hello!"}],
    )
    print(resp.text)

asyncio.run(main())

What's next?