Azure OpenAI Code Interpreter - Containers API - Timeout

Question

Azure OpenAI Code Interpreter - Containers API - Timeout

Akundy, Vyas 120

Hi Team,

I am trying to implement the Code Interpreter that is enabled through tool calling in the new responses API and as part of that to be able to perform Q&A tasks on a file the OpenAI documentation shows that we need to create a container and upload files to the container. However I am trying to run the following code and it always shows a timeout and does not list the containers that I have created. Is there any alternative that you can suggest here or let me know how to make this work?

from openai import OpenAI

client = OpenAI(
    api_key="api key",
    base_url="https://azure-openai-resource.openai.azure.com/openai/v1",
)
client.containers.list()

Output: {'error': {'code': 'Timeout', 'message': 'The operation was timeout.'}}

Akundy, Vyas 120 Reputation points

2025-11-11T20:40:22.8033333+00:00

Thanks for the response, just to confirm instead of tracking by container_ids I can use the returned file_ids to make sure when I ask a question to the model it will be able to pickup the correct file by ids even though I do not specify any particular container_id ?
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-11T20:46:18.8166667+00:00

Hi Akundy, Vyas

In Azure OpenAI’s Responses API with the Code Interpreter, you don’t need to track or specify container_id yourself. Instead, you just reference your uploaded file_ids, and Azure automatically handles container creation, association, and cleanup behind the scenes.
Akundy, Vyas 120 Reputation points

2025-11-12T14:59:34.2333333+00:00

Hi @Sridhar M , is there any recommended architecture around session tracking when it comes to containers and files. Because My assumption is that because Azure maintains lifecycle of a temporary container this would mean that if the container lifespan has ended then we would need to reupload the files once more to the container.Or is it the case that Once the files are uploaded via client.files.create() the files remain present permanently until they are deleted?
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T17:51:01.4766667+00:00

Hi Akundy, Vyas

“If the container lifespan has ended, do we need to reupload the files?” No, provided you have the file_ids. Just create a new response with Code Interpreter and include the existing file_ids in the container config; the service will attach them to the new container.

“Do files uploaded via client.files.create() remain permanently until deleted?” Yes, that’s the designed behavior they remain available in the Files store until you delete them (subject to your own retention/cleanup policies). They’re independent of any given container’s lifetime.
Akundy, Vyas 120 Reputation points

2025-11-12T18:07:05.3766667+00:00

Regarding the accepted file formats, I tried to upload an excel/csv file but it says they are not supported. Is there any date when we can expect excel file upload support would be added? This is the error I have received

{'error': {'code': 'invalidPayload', 'message':

'fileName contains an invalid filename: Unsupported extension: .xlsx. Supported

extensions: .txt, .md, .pdf, .json.'}}
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T18:13:54.62+00:00
Hi Akundy, Vyas

Currently, the Azure OpenAI file upload feature only supports the following extensions:

.txt

.md

.pdf

.json

Excel files (.xlsx) and CSV files are not supported yet, which is why you’re seeing the invalidPayload error.

Microsoft has not announced an official timeline for adding Excel or CSV support to the file upload API. These formats often require additional parsing and security checks, so they’re typically introduced later.

you can do now

Convert your Excel/CSV to a supported format:

Save as .txt or .json if the structure is simple.

For tabular data, export as .json for better compatibility.

Use preprocessing in your app:

Read the Excel/CSV locally using Python (pandas) or similar tools.

Convert the content into text or JSON before sending it to the API.
Akundy, Vyas 120 Reputation points

2025-11-12T20:07:11.7866667+00:00

Is there any charge for creating the container as I see something written here https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/code-interpreter?tabs=python but could not find any specific cost
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T21:16:49.65+00:00
Hi Akundy, Vyas

There is no separate charge for creating the container when using the Code Interpreter feature in Azure AI Foundry. The container is a temporary execution environment managed by the service, and its lifecycle is tied to your session.

Compute and storage usage behind the scenes is included in the overall pricing for the Azure OpenAI service.

You pay based on:

Tokens processed by the model (prompt + response).

Any additional resources you explicitly provision (e.g., persistent storage or external compute).

The documentation highlights containers because the Code Interpreter runs in an isolated environment for security and reproducibility. However, this does not introduce a separate SKU or cost line item.

Answer accepted by question author

0 additional answers

Your answer

Akundy, Vyas 120 Reputation points

2025-11-11T20:40:22.8033333+00:00

Thanks for the response, just to confirm instead of tracking by container_ids I can use the returned file_ids to make sure when I ask a question to the model it will be able to pickup the correct file by ids even though I do not specify any particular container_id ?
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-11T20:46:18.8166667+00:00

Hi Akundy, Vyas

In Azure OpenAI’s Responses API with the Code Interpreter, you don’t need to track or specify container_id yourself. Instead, you just reference your uploaded file_ids, and Azure automatically handles container creation, association, and cleanup behind the scenes.
Akundy, Vyas 120 Reputation points

2025-11-12T14:59:34.2333333+00:00

Hi @Sridhar M , is there any recommended architecture around session tracking when it comes to containers and files. Because My assumption is that because Azure maintains lifecycle of a temporary container this would mean that if the container lifespan has ended then we would need to reupload the files once more to the container.Or is it the case that Once the files are uploaded via client.files.create() the files remain present permanently until they are deleted?
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T17:51:01.4766667+00:00

Hi Akundy, Vyas

“If the container lifespan has ended, do we need to reupload the files?” No, provided you have the file_ids. Just create a new response with Code Interpreter and include the existing file_ids in the container config; the service will attach them to the new container.

“Do files uploaded via client.files.create() remain permanently until deleted?” Yes, that’s the designed behavior they remain available in the Files store until you delete them (subject to your own retention/cleanup policies). They’re independent of any given container’s lifetime.
Akundy, Vyas 120 Reputation points

2025-11-12T18:07:05.3766667+00:00

Regarding the accepted file formats, I tried to upload an excel/csv file but it says they are not supported. Is there any date when we can expect excel file upload support would be added? This is the error I have received

{'error': {'code': 'invalidPayload', 'message':

'fileName contains an invalid filename: Unsupported extension: .xlsx. Supported

extensions: .txt, .md, .pdf, .json.'}}
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T18:13:54.62+00:00

Hi Akundy, Vyas

Currently, the Azure OpenAI file upload feature only supports the following extensions:

.txt

.md

.pdf

.json

Excel files (.xlsx) and CSV files are not supported yet, which is why you’re seeing the invalidPayload error.

Microsoft has not announced an official timeline for adding Excel or CSV support to the file upload API. These formats often require additional parsing and security checks, so they’re typically introduced later.

you can do now

Convert your Excel/CSV to a supported format:

Save as .txt or .json if the structure is simple.

For tabular data, export as .json for better compatibility.

Use preprocessing in your app:

Read the Excel/CSV locally using Python (pandas) or similar tools.

Convert the content into text or JSON before sending it to the API.
Akundy, Vyas 120 Reputation points

2025-11-12T20:07:11.7866667+00:00

Is there any charge for creating the container as I see something written here https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/code-interpreter?tabs=python but could not find any specific cost
Sridhar M 3,160 Reputation points Microsoft External Staff Moderator

2025-11-12T21:16:49.65+00:00

Hi Akundy, Vyas

There is no separate charge for creating the container when using the Code Interpreter feature in Azure AI Foundry. The container is a temporary execution environment managed by the service, and its lifecycle is tied to your session.

Compute and storage usage behind the scenes is included in the overall pricing for the Azure OpenAI service.

You pay based on:

Tokens processed by the model (prompt + response).

Any additional resources you explicitly provision (e.g., persistent storage or external compute).

The documentation highlights containers because the Code Interpreter runs in an isolated environment for security and reproducibility. However, this does not introduce a separate SKU or cost line item.

Answer 1

Hi Akundy, Vyas

Welcome to Microsoft Q&A and Thank you for reaching out.

You’re seeing the Timeout Error because the Azure OpenAI Containers API isn’t fully exposed for direct management through the Python SDK. In Azure, the container feature is mainly an internal mechanism that supports the Code Interpreter tool (used within the Responses API) rather than a standalone service you can list or query directly.

So, when you call client.containers.list(), the SDK is trying to reach an endpoint (GET /openai/v1/containers) that either doesn’t exist in Azure’s implementation or isn’t enabled in your region hence the timeout.

In Azure OpenAI, you don’t actually need to manually create or list containers to use the Code Interpreter. Instead, Azure provides an “auto” mode where containers are automatically managed behind the scenes. You simply upload files with client.files.create() and then reference those file IDs in the Responses API request by specifying a container parameter with "type": "auto". This automatically creates a temporary container for your code execution task, which is valid for a limited time (usually around one hour with an idle timeout of 20 minutes).

If you still want to confirm whether container management endpoints are available, you can check Azure’s REST API documentation or try calling GET https://<your-resource>.openai.azure.com/openai/v1/containers directly with tools like Postman but most users report this endpoint is either restricted or unsupported in Azure.

In short, the timeout isn’t caused by your setup but by the fact that Azure OpenAI doesn’t yet support explicit container listing. The correct approach is to rely on automatic container creation via the Responses API using "container": {"type": "auto"} and file_ids.

References

Azure OpenAI Responses API overview & regions: [Azure Open...soft Learn | Learn.Microsoft.com]
Guidance on container file download URLs on Azure: [stackoverflow.com]
Official sample repo for Azure Responses API: [github.com]

I Hope this helps. Do let me know if you have any further queries.

Thank you!

Share via

Azure OpenAI Code Interpreter - Containers API - Timeout

0 additional answers

Your answer