Streaming¶

When stream=True, create() returns a ResponseStream that yields events as they arrive from the API.

Basic Streaming¶

with client.responses.create(
    model="gpt-5.1-codex-mini",
    instructions="Be helpful.",
    input="Tell me a joke.",
    stream=True,
) as stream:
    for event in stream:
        if isinstance(event, codex_open_client.ResponseOutputTextDeltaEvent):
            print(event.delta, end="", flush=True)
    print()

Getting the Final Response¶

You can consume the stream and get the complete Response:

stream = client.responses.create(
    model="gpt-5.1-codex-mini",
    instructions="Be brief.",
    input="Hello!",
    stream=True,
)

response = stream.get_final_response()
print(response.output_text)

Re-Iterating¶

A consumed stream caches its events. You can iterate again:

stream = client.responses.create(..., stream=True)

# First pass — print deltas
for event in stream:
    if isinstance(event, codex_open_client.ResponseOutputTextDeltaEvent):
        print(event.delta, end="")

# Second pass — count events (uses cached events, no network)
event_count = sum(1 for _ in stream)

Event Types¶

All events are dataclasses with a type field. Use isinstance() to filter:

Text Events¶

Event	Key Fields	Description
`ResponseOutputTextDeltaEvent`	`delta: str`	A chunk of text output
`ResponseOutputTextDoneEvent`	`text: str`	Complete text for one content block

Response Lifecycle¶

Event	Key Fields	Description
`ResponseCreatedEvent`	`response`	Response object created
`ResponseInProgressEvent`	`response`	Model is generating
`ResponseCompletedEvent`	`response`	Response finished successfully
`ResponseFailedEvent`	`response`	Response failed
`ResponseIncompleteEvent`	`response`	Response was cut short

Output Items¶

Event	Key Fields	Description
`ResponseOutputItemAddedEvent`	`item`, `output_index`	New output item started
`ResponseOutputItemDoneEvent`	`item`, `output_index`	Output item completed

Tool Calls¶

Event	Key Fields	Description
`ResponseFunctionCallArgumentsDeltaEvent`	`delta`	Chunk of function arguments
`ResponseFunctionCallArgumentsDoneEvent`	`arguments`	Complete function arguments

Reasoning¶

Event	Key Fields	Description
`ResponseReasoningSummaryTextDeltaEvent`	`delta`	Chunk of reasoning summary
`ResponseReasoningSummaryTextDoneEvent`	`text`	Complete reasoning summary

See Stream Events API Reference for full type details.

Context Manager¶

ResponseStream supports with statements to ensure the underlying HTTP response is closed:

with client.responses.create(..., stream=True) as stream:
    for event in stream:
        ...
# HTTP response is closed here