Your first call, in two minutes.

Keys live under a project, in Gateway → Keys. Create one, then copy it — the full secret is shown once, and the console only keeps its prefix afterwards. Every key carries the project it was minted in, which is how usage and billing get attributed.

Nothing below hard-codes the key, so the same commands are safe to paste into a shared terminal or a script you commit.

export LEGIONEDGE_API_KEY="lek_live_…"

One POST to /chat/completions with a model id and a list of messages. This runs on the shared pool — there is no GPU to start and nothing to wait for.

curl https://inference.legionedge.ai/v1/chat/completions \
  -H "Authorization: Bearer $LEGIONEDGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen3-4b-instruct", "messages": [{"role": "user", "content": "Hello!"}]}'

You get back the usual OpenAI response envelope: a choices[0].message.content with the reply, and a usage block whose token counts are exactly what the request is billed on.

The official OpenAI clients work unmodified — only base_url changes.

import os

from openai import OpenAI

client = OpenAI(
    base_url="https://inference.legionedge.ai/v1",
    api_key=os.environ["LEGIONEDGE_API_KEY"],
)

response = client.chat.completions.create(
    model="qwen3-4b-instruct",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)