Use the gateway

The gateway speaks the OpenAI chat-completions dialect. Any client, SDK, or script that works against OpenAI's API works here — swap the base URL and key, nothing else.

Endpoint

https://gw.theazizi.space/v1/chat/completions

Authentication uses your gateway key (azg_…) as a Bearer token. Streaming works — set "stream": true and read server-sent events exactly as you would with OpenAI.

Quickstart — curl

curl https://gw.theazizi.space/v1/chat/completions \
  -H "Authorization: Bearer azg_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-oss-20b:free",
    "messages": [
      { "role": "user", "content": "Explain failover in one sentence." }
    ]
  }'

Python (openai SDK)

from openai import OpenAI

client = OpenAI(
    base_url="https://gw.theazizi.space/v1",
    api_key="azg_YOUR_KEY",
)

resp = client.chat.completions.create(
    model="deepseek/deepseek-r1",
    messages=[{"role": "user", "content": "hello"}],
)
print(resp.choices[0].message.content)

Models

GET https://gw.theazizi.space/v1/models
Authorization: Bearer azg_YOUR_KEY

Model ids follow OpenRouter naming — browse the catalog at openrouter.ai/models and send any id.

Errors

401Missing or invalid gateway key.

403Key disabled or expired — contact the admin.

429Rate limit or daily token budget hit. Retry after a minute, or tomorrow for daily limits.

502Every pool key failed on this request — usually transient.

503No healthy keys in the pool right now.