Send your first request
Use the default project, create an API key, and make an authenticated request.
1. Use the default project
Sign in to the Star OS Console and select Default Project to get started immediately. You can also create your own projects to separate applications, teams, or environments.
Your API keys, billing, and model profiles are all bound to a project. This keeps credentials, model configuration, usage, and charges together in the same workspace.
2. Create an API key
Open API Keys in your project, select Create API Key, and copy the secret immediately. Treat it as a password: do not commit it to source control or expose it in browser code.
3. Store the key securely
For a server-side application, keep the key in an environment variable.
export ARK_API_KEY="your-api-key"4. Choose a model ID
Every inference request must include a model parameter. Set it to the Model ID of a model available to your project—not its display name.
In the Star OS Console, open Models, find a model with the Available status, and copy the value in the Model ID column. Use that exact value in your request:
{
"model": "your-model-id",
"messages": [{ "role": "user", "content": "Hello!" }]
}Model availability and the model profiles your application can use are configured per project. If a model ID is unavailable in your project, choose an available model from the Console. See text generation models and pricing for the current Qwen, GLM, and DeepSeek catalog.
5. Make a request
The public inference API is provided by Ark Gateway. Its published base URL and supported request shapes will be added to this guide together with the public Gateway OpenAPI contract.
Do not use the Dashboard Product API or Gateway administration endpoints for inference traffic.
Next steps
- Read Projects and API keys.
- Configure the models available to your project in the Console.
- Review usage and billing before enabling production traffic.