Prepare Azure before RapidClaw.

Create a new Azure subscription for RapidClaw so the runtime, quota, cost, access, and policy boundary are easy to understand.

The subscription should belong to the customer tenant and be used specifically for RapidClaw. This keeps the deployment isolated from unrelated Azure workloads and makes billing, governance, and support conversations cleaner.

• Create a new customer-owned Azure subscription for RapidClaw.
  Open Azure Subscriptions
  If you cannot access this link, you will need higher permissions to deploy RapidClaw.
• Use a clear name, such as RapidClaw Production or RapidClaw Pilot.
• Strongly recommended: Set a budget Alert.
• Confirm the subscription is active.

RapidClaw needs available model capacity in the subscription and region selected for deployment.

A valid Azure subscription is not enough by itself. Azure AI Foundry / Azure OpenAI capacity is quota-based, and quota availability varies by region and model family.

Open the official Microsoft quota instructions so you can confirm the current Microsoft process.
Open Microsoft quota instructions

• Open Azure AI Foundry.
• Confirm you are signed in to the correct customer tenant. If you are not, switch directories before continuing.
• Open or create a Foundry project tied to the dedicated RapidClaw subscription.
• In the upper navigation, select Operate.
• In the left pane, select Quota.
• Use the quota filters or table to select the dedicated RapidClaw subscription.
• Select the same Azure region planned for RapidClaw deployment.
• Find the required Azure OpenAI / Foundry model family and deployment type for RapidClaw.
• Select Request quota and enter the requested capacity for the pilot or production workload.
• Submit the request, then wait for Microsoft approval before scheduling RapidClaw deployment. Approval may take a day or two.

2.1Recommended quota request form values

For the Azure OpenAI quota request form, use these RapidClaw recommendations:

• 11. Model Type: Azure OpenAI
• 12. Model Deployment Quota or Fine Tuning Quota: Model Deployment
• 13. (Azure OpenAI) Quota Request Type: Global Standard
• 14. (Azure OpenAI) Global Standard Model: gpt-5.5
• 15. Global Standard Region: Select the same region planned for RapidClaw deployment.
• 16. (Azure OpenAI) Global Standard Quota: Enter 100. Do not enter 100,000. This field is measured in kTPM, so entering 100 represents 100,000 tokens per minute total quota for the initial RapidClaw deployment.

Quota is a capacity limit, not a usage charge. You do not pay for the requested quota itself. Azure OpenAI / Foundry billing is based on tokens actually used by RapidClaw.