Python: [Bug]: Foundry Hosted agent - delay before final response when streaming

[cecheta]

Description

When a Foundry Hosted agent streams a response, there is a ~5 delay between the last response.output_item.done item and the final response.completed event. This can be seen in the logs:

2026-08-03 14:15:02,826 INFO azure.ai.agentserver: Inbound POST /responses started
2026-08-03 14:15:02,827 INFO azure.ai.agentserver: Creating response caresp_205688e047e259e500PbZ66Mj3R6Omb1FaSrJUaK3conqWDk9c: streaming=True background=False store=True model= conversation_id=None previous_response_id=None has_user_id=False has_call_id=False
2026-08-03 14:15:02,828 INFO azure.ai.agentserver: Invoking handler ResponsesAgentServerHost._dispatch_create for response caresp_205688e047e259e500PbZ66Mj3R6Omb1FaSrJUaK3conqWDk9c
2026-08-03 14:15:02,830 INFO agent_framework: {'role': <MessageRole.USER: 'user'>, 'parts': [{'type': 'text', 'content': 'Hi'}]}
2026-08-03 14:15:02,830 INFO agent_framework: {'role': <MessageRole.USER: 'user'>, 'parts': [{'type': 'text', 'content': 'Hi'}]}
2026-08-03 14:15:03,628 INFO httpx: HTTP Request: POST https://aif-7f6zzbbgxhhmy.services.ai.azure.com/api/projects/proj-7f6zzbbgxhhmy/openai/v1/responses "HTTP/1.1 200 OK"
2026-08-03 14:15:04,155 WARNING agent_framework_foundry_hosting._responses: Content type 'usage' is not supported yet. This is usually safe to ignore.
2026-08-03 14:15:09,171 INFO agent_framework: {'role': 'assistant', 'parts': [{'type': 'text', 'content': 'Hi! How can I help?'}]}
2026-08-03 14:15:09,172 INFO agent_framework: {'role': 'assistant', 'parts': [{'type': 'text', 'content': 'Hi! How can I help?'}]}
2026-08-03 14:15:09,176 INFO azure.ai.agentserver: 127.0.0.1:37214 "POST /responses 1.1" 200 - 6349804μs
2026-08-03 14:15:09,177 INFO azure.ai.agentserver: Inbound POST /responses completed with status 200 in 6350.2ms

Code Sample

import os

from agent_framework import Agent
from agent_framework.foundry import FoundryChatClient
from agent_framework_foundry_hosting import ResponsesHostServer
from azure.identity import DefaultAzureCredential


def main():
    model_name = os.getenv("AZURE_AI_MODEL_DEPLOYMENT_NAME") or os.getenv("FOUNDRY_MODEL_NAME")
    if not model_name:
        raise RuntimeError(
            "Model deployment name is not configured. Set AZURE_AI_MODEL_DEPLOYMENT_NAME or FOUNDRY_MODEL_NAME."
        )

    client = FoundryChatClient(
        project_endpoint=os.environ["FOUNDRY_PROJECT_ENDPOINT"],
        model=model_name,
        credential=DefaultAzureCredential(),
    )

    agent = Agent(
        client=client,
        instructions="You are a friendly assistant. Keep your answers brief.",
        default_options={"store": False},
    )

    server = ResponsesHostServer(agent)
    server.run()


if __name__ == "__main__":
    main()

Error Messages / Stack Traces

Package Versions

agent-framework-core:1.13.0, agent-framework-foundry: 1.10.4, agent-framework-foundry-hosting: 1.0.0b260730, agent-framework-openai: 1.12.0

Python Version

Python 3.13

Additional Context

No response

[github-actions]

👋 Hi @cecheta — thanks for the report!

Your logs clearly show the ~5 second gap between 14:15:04 (last content event) and 14:15:09 (response completion), which is consistent with the delay you describe. Unfortunately, our automated investigation hit infrastructure issues and could not inspect the current codebase to verify whether this has already been addressed. Could you confirm whether you're still seeing this delay with the latest version of agent-framework-foundry-hosting?

Once you can share the details above, we'll take another look. 🙏

[cecheta]

👋 Hi @cecheta — thanks for the report!

Your logs clearly show the ~5 second gap between 14:15:04 (last content event) and 14:15:09 (response completion), which is consistent with the delay you describe. Unfortunately, our automated investigation hit infrastructure issues and could not inspect the current codebase to verify whether this has already been addressed. Could you confirm whether you're still seeing this delay with the latest version of agent-framework-foundry-hosting?

Once you can share the details above, we'll take another look. 🙏

Yes, 1.0.0b260730

[cecheta]

Seems like this is caused by a combination of:

• Using the Microsoft Foundry project endpoint (https://<RESOURCE>.services.ai.azure.com/api/projects/<PROJECT>/openai/v1)
• Setting "store": False

So it is not caused by the SDK.