Quickstart
Mint a key, send a request, and see it land in usage — about two minutes end to end.
You need an account with credit on it and a terminal. Nothing to install.
1. Mint a key
Open API Keys and create one. The raw key is shown once and never again — there is no recovery path, which is the point. Copy it somewhere safe.
Keys look like this:
sk-ww-3f8c1d0a94b7e25610fa8c3d47b9e0125faa6d38c1b0947e
Only the first ten characters are stored in a readable form, so the dashboard
can show you sk-ww-3f8c and nothing more.
2. Send a request
curl -sS "https://api.webway.example/v1/chat/completions" \
-H "Authorization: Bearer $WEBWAY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-5.6-luna",
"messages": [{"role": "user", "content": "Hello!"}],
"max_tokens": 64
}'
The response is the shape an OpenAI client already expects, so any SDK that
takes a base_url works unchanged:
from openai import OpenAI
client = OpenAI(
base_url="https://api.webway.example/v1",
api_key=os.environ["WEBWAY_API_KEY"],
)
3. Watch it land
Open Usage. The request appears as part of the day's totals within a minute of finishing — tokens in, tokens out, what it cost, and how long the first token took.
If the spend column reads $0.00, the call was cheap rather than free. Costs
are held in micro-dollars and only rounded when they are printed.
Where next
- Your first request — the body, field by field.
- Authentication — key handling and rotation.
- Streaming — token-by-token responses.