Azure Cognitive Services and Standard Text Records as additional charged to my billing account

Question

Hey guys, I just want to ask why am I charged for an Azure Cognitive Services and Standard Text Records even if I only deployed Phi-4-mini-instruct and text-embedding-3-small in my foundry project? I also wanted to find that service in my portal but I'm having a hard time looking for it.

Lastly, I wanted to ask for explanations what are those Azure Cognitive Services for and how is it relevant to my deployed models? Can I disable it and be charged only on the 2 deployed models?

Answer 1

Hello Charles Bernard Balaguer,

Greetings! Thanks for raising this question in Q&A forum.

The most likely reason is that Azure AI Foundry model usage can generate charges that appear under Azure Cognitive Services billing meters rather than showing only the model names themselves. The Standard Text Records charge is often associated with text processing operations performed by deployed AI models and related Azure AI services.

1. Go to Cost Management + Billing and review the detailed usage records for the billing period to identify which resource and meter generated the charges.
2. Check the resource group used by your AI Foundry project and verify whether an Azure AI Services/Cognitive Services resource was automatically created or linked during project setup.
3. Review the usage of your Phi-4-mini-instruct and text-embedding-3-small deployments, as requests to these models may be billed through Cognitive Services meters.
4. In the Azure portal, use Cost Analysis and group costs by Resource and Meter to determine exactly which service is generating the "Standard Text Records" charges.
5. If the charges are unclear after reviewing the billing details, open a billing support request and provide the subscription ID and affected billing period so the billing team can verify the specific meter usage.

The next action is with the customer. The expected outcome is to identify the exact resource and billing meter behind the charge. If the charge is tied to model inference or embedding requests, it generally cannot be disabled separately from the model usage itself. Microsoft Billing Support can confirm the exact mapping between the meter and your deployed models if needed.

If this answer helps you kindly accept the answer which will help others who have similar questions

Best Regards,

Jerald Felix.