Introduction
AI in ARVO Studio runs on the customer’s own Azure AI Foundry deployment. That deployment sits in the customer’s Azure subscription, under their own agreement with Microsoft, and Microsoft bills them for what it consumes. This article covers the fields that connect ARVO Studio to the deployment, and the Azure work needed to create the deployment in the first place.
Use this article if you are connecting through Azure AI Foundry. If you are using your own Anthropic account instead, see Your Anthropic API (BYO) article.
Note: AI runs on the customer’s own Azure AI Foundry deployment or Anthropic account and is paid for by the customer. Bluefort does not resell, supply or absorb AI capacity, and AI usage is not included in the ARVO fee.
Fields
Complete the following on ARVO Studio Setup, in the Azure AI Foundry group.
| Field | What to enter | Notes |
| Foundry Connection Enabled | Turn on to use Azure AI Foundry for ARVO Studio AI operations | Only one AI connection should be active at a time: Foundry or Anthropic direct. |
| Foundry Endpoint URL | The endpoint URL for your Foundry deployment | Use a placeholder resource name in screenshots, never a real customer endpoint. |
| Model Deployment | The deployment name created in Azure AI Foundry | The supported models are listed under Supported models and recommended quota in this article, which is the single maintained page for them. |
| Foundry Key and Key Set | The deployment key. Key Set confirms that a key has been stored. | Write-only. If the key is rotated in Foundry, enter the new key here. |
| LLM Max Tokens | Maximum output tokens per call | Keep the default unless the customer’s deployment enforces a lower limit. A value that is too low can truncate output and cause validation to fail. |
Only one AI connection at a time
Only one AI connection should be active at a time, either Azure AI Foundry or Anthropic direct. Decide which provider the customer is using, enable that connection and leave the other one switched off.
Create a Foundry deployment for ARVO Studio
This part of the article is written for the Azure administrator who prepares the deployment. It is a short piece of Azure work and it can be done before, or alongside, the rest of ARVO Studio Setup.
- Create an Azure AI Foundry resource in the customer’s subscription, in a supported region, or reuse an existing resource that the customer is content to use for this work.
- Deploy a supported model to that resource and give the deployment a name you will recognise later. The supported models are listed further down this article, under Supported models and recommended quota.
- Copy the endpoint URL and a key from the deployment and pass them to whoever is completing ARVO Studio Setup, using the customer’s normal method for handling secrets.
- Check the quota assigned to the deployment before the first build, so that work is not throttled part-way through. The recommended quota is given further down this article.
Supported models and recommended quota
This article is the single maintained page for the supported model list and the recommended quota. Model names are not repeated anywhere else in the portal, so that there is only one page to update when model support changes, and every other article links here instead.
Quota, throttling, latency and regional availability are controlled by Microsoft, not by Bluefort. If work is slowed or refused because of a limit on the deployment, the change has to be made in the customer’s own Azure subscription.
Quick check
Save the group and confirm that the Access Key configured flag shows that a key is stored. Then run Test Connection, which reports whether the selected AI connection responds. If it does not, check the endpoint URL and the deployment name first, then re-enter the key, and then look at the quota on the deployment.