OpenAI Studio: "The specified embedding deployment 'text-embedding-ada-002' in the request body does not exist."

Question

OpenAI Studio: "The specified embedding deployment 'text-embedding-ada-002' in the request body does not exist."

Maria Natalia Benavides Sanabria 10

Dear Tem,

error (2)

I am experiencing the following error in Azure OpenAI Studio: "The specified embedding deployment 'text-embedding-ada-002' in the request body does not exist." This error appears when I add the option to perform vector search while testing with my own data, even though I have already created the text embedding model.
implementacion

I am using GPT-4, Azure Blob Storage, and the AI search service. Could you help me understand why this error occurs and how I can resolve it?
pasos

Thank you for your assistance.

YutongTie-9091 54,031 Reputation points Moderator

2024-06-06T00:20:45.6733333+00:00

Hello, thanks for reaching out to us, we are checking this issue internally. At the meantime, could you please try the latest version of model to see if that works? -

text-embedding-3-large is the latest and most capable embedding model.
Ruth Shangobiyi 0 Reputation points Microsoft Employee

2024-06-23T17:59:27.6266667+00:00

@Anonymous Is there an update?
Raymond Lee 10 Reputation points

2025-08-27T06:19:17.69+00:00

I ran into a similar situation. I am self-teaching myself AI fundamentals and I am following the tutorial Develop a RAG-based solution with your own data using Azure AI Foundry.

The same error occurred for me, where the embedding deployment "insert_embedding_model_name" did not exist. I was able to resolve it by ensuring that both my gpt-4o model and my embedding (ada_002) model were deployed in the same AI resource location.

Before, I had gpt-4o deployed on useast, and ada_002 on useast2.

Once I had deployed a new embedding model in the same resource location, my prompt was successfully able to use the indexed data (hybrid search).

I hope this helps anyone running into the same issue.

Note: I was able to recreate the problem if I created the embedding model first, before gpt-4o. If so, I wouldn't be able to select the same resource location due to "no quota available". It seems like the solution is to create the gpt-4o model first, then create the embedding model. This is what allowed me to select the same location for both models.
RAMIA ALI 5 Reputation points

2025-10-03T15:05:18.9666667+00:00

Hi Raymond
Thank you. I had the same issue and could resolve the problem by deploying the embedding model on the same location as gpt-4o (using the same ai ressource -> same location)

Your answer

YutongTie-9091 54,031 Reputation points Moderator

2024-06-06T00:20:45.6733333+00:00

Hello, thanks for reaching out to us, we are checking this issue internally. At the meantime, could you please try the latest version of model to see if that works? -

text-embedding-3-large is the latest and most capable embedding model.
Ruth Shangobiyi 0 Reputation points Microsoft Employee

2024-06-23T17:59:27.6266667+00:00

@Anonymous Is there an update?
Raymond Lee 10 Reputation points

2025-08-27T06:19:17.69+00:00

I ran into a similar situation. I am self-teaching myself AI fundamentals and I am following the tutorial Develop a RAG-based solution with your own data using Azure AI Foundry.

The same error occurred for me, where the embedding deployment "insert_embedding_model_name" did not exist. I was able to resolve it by ensuring that both my gpt-4o model and my embedding (ada_002) model were deployed in the same AI resource location.

Before, I had gpt-4o deployed on useast, and ada_002 on useast2.

Once I had deployed a new embedding model in the same resource location, my prompt was successfully able to use the indexed data (hybrid search).

I hope this helps anyone running into the same issue.

Note: I was able to recreate the problem if I created the embedding model first, before gpt-4o. If so, I wouldn't be able to select the same resource location due to "no quota available". It seems like the solution is to create the gpt-4o model first, then create the embedding model. This is what allowed me to select the same location for both models.
RAMIA ALI 5 Reputation points

2025-10-03T15:05:18.9666667+00:00

Hi Raymond
Thank you. I had the same issue and could resolve the problem by deploying the embedding model on the same location as gpt-4o (using the same ai ressource -> same location)