openai/openai-python · warning · RuntimeError

No run steps found

Error message

No run steps found

What it means

Raised by AssistantEventHandler.get_final_run_steps after stream completion when __run_step_snapshots is empty. Run steps are accumulated from thread.run.step.* events; if none arrived (the run produced no tool calls/details, or the stream never delivered run-step events), the dict stays empty and this error is thrown.

Source

Thrown at src/openai/lib/streaming/_assistants.py:122

    def until_done(self) -> None:
        """Waits until the stream has been consumed"""
        consume_sync_iterator(self)

    def get_final_run(self) -> Run:
        """Wait for the stream to finish and returns the completed Run object"""
        self.until_done()

        if not self.__current_run:
            raise RuntimeError("No final run object found")

        return self.__current_run

    def get_final_run_steps(self) -> list[RunStep]:
        """Wait for the stream to finish and returns the steps taken in this run"""
        self.until_done()

        if not self.__run_step_snapshots:
            raise RuntimeError("No run steps found")

        return [step for step in self.__run_step_snapshots.values()]

    def get_final_messages(self) -> list[Message]:
        """Wait for the stream to finish and returns the messages emitted in this run"""
        self.until_done()

        if not self.__message_snapshots:
            raise RuntimeError("No messages found")

        return [message for message in self.__message_snapshots.values()]

    def __text_deltas__(self) -> Iterator[str]:
        for event in self:
            if event.event != "thread.message.delta":
                continue

            for content_delta in event.data.delta.content or []:

View on GitHub (pinned to 9917c6e28e)

Solutions

1. Treat empty steps as valid: wrap in try/except RuntimeError and use an empty list
2. If you override event callbacks in a subclass, call super().on_event(event) (or accumulate_event) so snapshots are recorded
3. Verify with get_final_run() that the run actually completed successfully
4. Check that tool use was actually expected for this run

Example fix

# before
steps = stream.get_final_run_steps()

# after
try:
    steps = stream.get_final_run_steps()
except RuntimeError:
    steps = []  # run had no tool steps

null

null

try:
    steps = stream.get_final_run_steps()
except RuntimeError:
    steps = []  # no tool steps in this run

When it happens

Trigger: Calling get_final_run_steps() on a simple run that only produced a text message (no tool steps), or on a stream that terminated before any thread.run.step.* events were emitted; also when custom on_event overrides skip the accumulate_event call.

Common situations: Assistants v2 streaming where the model answered directly with no tools; custom handler subclasses overriding on_event without calling super(); interrupted or cancelled runs.

Related errors

• A single event handler cannot be shared between multiple str
• No final run object found
• No messages found
• Stream has not been started yet
• Encountered a message delta with no previous snapshot

AI-assisted analysis of openai/openai-python@9917c6e28e (2026-08-28). Data as JSON: /api/errors/59b3f471b26e9115. Report an issue: GitHub.