Less guesswork.
A clearer first step.
A practical guide to planning your integration. Confirm access and configuration with our team before sending production traffic.
Integration preview, not a live API reference. This static site does not provision resources, issue API keys or run inference. Examples illustrate the intended request shape. Confirm the endpoint, protocol support and model IDs during onboarding.
01 / Start with your workload.
Share your chosen model or GPU, application use case, expected traffic, region requirements and target launch date. We’ll confirm what is available and what it takes to get started.
02 / Confirm your configuration.
Your onboarding details should include the following. Product names in the pricing catalog are display names, not executable API model IDs. Confirm the exact model release, capabilities and applicable license before deployment.
| Setting | What to confirm |
|---|---|
ARTRUBY_BASE_URL | Your assigned API base URL, including any version prefix. No live endpoint is published in this preview. |
ARTRUBY_API_KEY | Your authorized credentials. Keep them server-side and out of client code or public repositories. |
YOUR_ASSIGNED_MODEL_ID | The exact model identifier, version and permitted use. |
| Capabilities & limits | Input formats, context and output limits, streaming, tools, rate limits and concurrency. |
| Region & data handling | Processing location, retention scope and service-specific terms. |
03 / The shape of a first request.
For an endpoint confirmed to support chat completions, replace the model placeholder with the exact ID assigned to you. Set your base URL and key in your local environment; do not enter secrets into this website.
# Use the endpoint & model ID from onboarding
curl "$ARTRUBY_BASE_URL/chat/completions" \
-H "Authorization: Bearer $ARTRUBY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "YOUR_ASSIGNED_MODEL_ID",
"messages": [{
"role": "user",
"content": "Let’s build something great."
}]
}' This example does not send a request when viewed or copied. Features such as streaming and tool calls are model- and deployment-dependent; verify them before relying on them.
04 / Understand what you’re measuring.
For API planning, separate standard input, cached input and output tokens. Use cache rates only for tokens reported as eligible cache hits.
Estimated cost = uncached input × input rate + cached input × cache rate + output × output rate. Token quantities are expressed in millions.
05 / Before you ship.
- Verify the assigned model ID, supported input types and response format.
- Confirm rate limits, billing units, region and availability terms.
- Test timeout handling, bounded retries and your application’s fallback behavior.
- Review the service-specific data-handling and access-control requirements.
- Agree on a support contact and escalation process.
Read the Data Policy and Terms of Service, and confirm that they cover your deployment.
The next thing you build
starts with the right foundation.
Tell us what you’re working on. We’ll help you find the right fit.