Create a key for your application

Do not give an application the platform master key. Create a separate key that you can limit, replace, or revoke without affecting the rest of DeltaLLM.

Before you start

Every application key belongs to a team. If you do not have a team yet, create an organization and then a team.

Use the addresses that match your setup:

Setup	Admin UI	API
Docker	http://localhost:4002	http://localhost:4002
Development setup	http://localhost:5000	http://localhost:8000

Create the key

1. Open the Admin UI address for your setup and sign in.
2. Open API Keys.
3. Select Create key.
4. Choose the team that owns the application.
5. Keep the owner set to You, or choose a service account for a shared application.
6. Give the key a clear name, such as support-app-local.
7. Add a budget, expiry date, or rate limit if the application needs one.
8. Choose which models or access groups the application may use.
9. Create the key and copy it immediately. The full key is shown only once.

Store the key in a secret manager or environment variable. Do not commit it to source control.

Test the key

curl http://localhost:4002/v1/models \
  -H "Authorization: Bearer YOUR_APPLICATION_KEY"

The command uses the Docker API address. For the development setup, replace it with http://localhost:8000. The response lists the models available to this key.

Connect an application

DeltaLLM works with OpenAI SDKs. Use the application key you just created and the public model name from the model list.

Install the SDK in your application project:

PythonJavaScript

python3 -m pip install openai

npm install openai

Then send a request. These examples use the Docker API address; use http://localhost:8000/v1 for the development setup.

PythonJavaScript

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:4002/v1",
    api_key="YOUR_APPLICATION_KEY",
)

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

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "http://localhost:4002/v1",
  apiKey: "YOUR_APPLICATION_KEY",
});

const response = await client.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(response.choices[0].message.content);

Replace gpt-4o-mini in either example if you chose a different public model name.

See Application keys for ownership, budgets, rate limits, and model access. See the admin API reference if you want to create keys automatically.