OpenAI Responses API calls + code interpreter tool have been failing (hanging ) in the past week or so

Question

OpenAI Responses API calls + code interpreter tool have been failing (hanging ) in the past week or so

GS 380

We are starting to see more and more OpenAI responses API calls that have the code interpreter tool never return .. in the past week or so 1 out of 4 in general fall into this category.

I am able to reproduce with Postman.

Some calls do succeed but it is becoming concerning and I am not too sure what the issue is exactly .. is anyone else experiencing this ?
I am seeing this in my dev instance and production instance .. both in the same tenant

I am setting the container type to auto

"tools":[{"type":"code_interpreter","container":{"type":"auto"}}]

Brian Muray 5 Reputation points

2026-02-24T19:13:35.7966667+00:00

I have the same issue as above, when the code_interpreter is enabled its minutes before I receive the first streamed output of response.created. When I remove the code_interpreter tool, I get a response in sub 2 seconds time.
Jerald Felix 10,975 Reputation points

2026-02-25T04:09:16.77+00:00

I am sorry to know that my answer not helped you will check and share you a better solution
GS 380 Reputation points

2026-02-25T09:22:58.9366667+00:00

Your response has nothing to do with my problem. You should delete it as to not mislead users.
AbP 10 Reputation points

2026-02-25T15:21:10.55+00:00

Having the same issue.
With "tools":[{"type":"code_interpreter","container":{"type":"auto"}}], /response api takes forever(over 5 mins) to respond, if at all. but mostly fails.

Trying to manually create containers with /containers rarely works or times out or takes over 5 mins.

https://<resource>.openai.azure.com/openai/v1/ endpoints. east us 2
GS 380 Reputation points

2026-02-25T16:30:02.9833333+00:00

@AbP Thank you for reporting that you are having the issue too - it is completely broken ..

Manas Mohanty 14,990 Microsoft External Staff Moderator

Hi GS

Could you share the region and sample code and point out any presence of VNET.

I am able to get results in Sweden Central region with below code.

import os
from openai import OpenAI

client = OpenAI(  
  base_url = "https://<openainame>.openai.azure.com/openai/v1/",
  api_key="<apikey>"  
)

instructions = "You are a personal math tutor. When asked a math question, write and run code using the python tool to answer the question."

response = client.responses.create(
    model="gpt-4.1-mini",
    tools=[
        {
            "type": "code_interpreter",
            "container": {"type": "auto"}
        }
    ],
    instructions=instructions,
    input="I need to solve the equation 3x + 11 = 14. Can you help me?",
)

print(response.output)

[ResponseOutputMessage(id='msg_0d404a3f6b36b42400699f5b2d00e8819591e7d07b0f4fae0d', content=[ResponseOutputText(annotations=[], text='Certainly! The equation to solve is:\n\n3x + 11 = 14...

Thank you.

GS 380 Reputation points

2026-02-25T20:45:57.5933333+00:00

Region East US 2 .

I am able to get responses too sometimes, but a lot of times things are timing out.

It has been that way since last week and this week.

It is intermittent.

2 answers

Your answer

Brian Muray 5 Reputation points

2026-02-24T19:13:35.7966667+00:00

I have the same issue as above, when the code_interpreter is enabled its minutes before I receive the first streamed output of response.created. When I remove the code_interpreter tool, I get a response in sub 2 seconds time.
Jerald Felix 10,975 Reputation points

2026-02-25T04:09:16.77+00:00

I am sorry to know that my answer not helped you will check and share you a better solution
GS 380 Reputation points

2026-02-25T09:22:58.9366667+00:00

Your response has nothing to do with my problem. You should delete it as to not mislead users.
AbP 10 Reputation points

2026-02-25T15:21:10.55+00:00

Having the same issue.
With "tools":[{"type":"code_interpreter","container":{"type":"auto"}}], /response api takes forever(over 5 mins) to respond, if at all. but mostly fails.

Trying to manually create containers with /containers rarely works or times out or takes over 5 mins.

https://<resource>.openai.azure.com/openai/v1/ endpoints. east us 2
GS 380 Reputation points

2026-02-25T16:30:02.9833333+00:00

@AbP Thank you for reporting that you are having the issue too - it is completely broken ..
Manas Mohanty 14,990 Reputation points Microsoft External Staff Moderator

2026-02-25T20:41:05.9566667+00:00

Hi GS

Could you share the region and sample code and point out any presence of VNET.

I am able to get results in Sweden Central region with below code.

import os from openai import OpenAI client = OpenAI( base_url = "https://<openainame>.openai.azure.com/openai/v1/", api_key="<apikey>" ) instructions = "You are a personal math tutor. When asked a math question, write and run code using the python tool to answer the question." response = client.responses.create( model="gpt-4.1-mini", tools=[ { "type": "code_interpreter", "container": {"type": "auto"} } ], instructions=instructions, input="I need to solve the equation 3x + 11 = 14. Can you help me?", ) print(response.output)

[ResponseOutputMessage(id='msg_0d404a3f6b36b42400699f5b2d00e8819591e7d07b0f4fae0d', content=[ResponseOutputText(annotations=[], text='Certainly! The equation to solve is:\n\n3x + 11 = 14...

Thank you.
GS 380 Reputation points

2026-02-25T20:45:57.5933333+00:00

Region East US 2 .

I am able to get responses too sometimes, but a lot of times things are timing out.

It has been that way since last week and this week.

It is intermittent.

Answer 1

Manas Mohanty 14,990 Microsoft External Staff Moderator

Hi GS

Thank you for sharing the details on support ticket and resource id.

Concerned engineering ticket associated with support ticket is in mitigated state post remedial action were taken from product group.

Status - Service is normal at customer side as per last interaction in private message.

Thank you for your inputs and using Azure services.

Answer 2

divyansh agarwal 160

If your code_interpreter runs are hanging, here’s what you should try:

Implement proper polling — keep checking the run status until it reaches completed or failed. Don’t assume the initial call will return final output.
Add a client-side timeout + retry logic for runs stuck in queued or in_progress.
Log request IDs and timestamps (UTC) and check if the issue correlates with peak traffic times.
Test with smaller inputs to rule out long-running or blocking code execution.
If the issue persists (~25% failure rate), open a support ticket and include:
- Request IDs
- Model name
- Region
- Exact payload

If it’s happening in both dev and prod, it’s likely backend/container allocation behavior — not your tenant configuration

GS 380 Reputation points

2026-02-25T20:49:01.9933333+00:00

I switched to creating my own containers with the v1/containers api for that reason .. i am able to implement exponential back off - retries etc..

it was fine for a little bit but now i am running into the same timeouts which is annoying - I do have a support ticket open with Microsoft

Share via

OpenAI Responses API calls + code interpreter tool have been failing (hanging ) in the past week or so

2 answers

Your answer