OpenAIIntegration breaks litellm streaming — LegacyAPIResponse wraps Stream objects

[iRbouh]

How do you use Sentry?

Sentry SaaS (sentry.io)

Version

sentry-sdk 2.56.0

Steps to Reproduce

import sentry_sdk
import litellm

sentry_sdk.init(dsn="...")  # OpenAIIntegration auto-enables

# This fails:
response = litellm.completion(
    model="gpt-4.1-nano",
    messages=[{"role": "user", "content": "hello"}],
    stream=True,
)
for chunk in response:
    print(chunk)

Versions:

• sentry-sdk==2.56.0
• openai==2.30.0
• litellm==1.82.6

Expected Result

Streaming works normally. litellm iterates chunks from the OpenAI response.

Actual Result

litellm.InternalServerError: OpenAIException - 'LegacyAPIResponse' object has no attribute '_iterator'
LiteLLM Retried: 3 times

The OpenAIIntegration monkey-patches the openai client and wraps the streaming response in a LegacyAPIResponse object. This object lacks the _iterator attribute that litellm expects on Stream objects. The OpenAI API returns 200 OK every time — the error occurs when litellm tries to iterate the wrapped response.

Workaround:

from sentry_sdk.integrations.openai import OpenAIIntegration

sentry_sdk.init(
    dsn="...",
    disabled_integrations=[OpenAIIntegration],
)

Disabling only OpenAIIntegration fixes the issue. All other auto-enabled integrations (HttpxIntegration, LiteLLMIntegration, etc.) work fine.

[linear-code]

PY-2269

[alexander-alderman-webb]

Thanks a lot for the report @iRbouh! We'll get a fix out next week.