Skip to main content
What this builds. Two demonstrations of cancellation against a slow LLM pipeline: (1) handler.terminate() on the RunHandler returned by pipeline.start(...), and (2) pipeline.terminate(run_id=...) using the raw task id. You’ll end up with. Two terminated runs with their post-termination status visible, useful when wiring a user “Stop generating” button. 
import time

from vectorshift.pipeline import Pipeline

# Create an LLM pipeline (LLM calls take a few seconds, giving time to terminate)
PIPELINE_NAME = "terminate_example"
try:
    pipeline = Pipeline.fetch(name=PIPELINE_NAME)
    print(f"Pipeline fetched: id={pipeline.id}, branch_id={pipeline.branch_id}")
except Exception as e:
    print(f"Error fetching pipeline: {e}")
    pipeline = Pipeline.new(name=PIPELINE_NAME)
    print(f"Pipeline created: id={pipeline.id}, branch_id={pipeline.branch_id}")
inp = pipeline.add(name="input_0", id="input_0").input(input_type="string")
llm = pipeline.add(name="llm", id="llm").llm(
    provider="openai", model="gpt-4o", prompt=inp.text
)
out = pipeline.add(name="output_0", id="output_0").output(
    output_type="string", value=llm.response
)
pipeline.save(deploy=True)

# Start a background run and terminate via RunHandler
handler = pipeline.start(
    inputs={"input_0": "Write a 500 word essay about the history of computing."}
)
print(f"Started: {handler.task_id}")
time.sleep(2)

result = handler.terminate()
print(f"Terminated via handler: {result}")
print(f"Status after termination: {handler.run_status()}")

# Start another and terminate via Pipeline directly using a run_id
handler2 = pipeline.start(
    inputs={"input_0": "Write another 500 word essay about robotics."}
)
time.sleep(2)

result2 = pipeline.terminate(run_id=handler2.task_id)
print(f"Terminated via pipeline: {result2}")

​
Expected output

Pipeline created: id=..., branch_id=...
Started: ...
Terminated via handler: {...}
Status after termination: {'status': 'terminated', ...}
Terminated via pipeline: {...}
task_id is the same identifier you’d pass to pipeline.terminate(run_id=...) — handlers are just a thin wrapper around it.

​
See also

Background runs

Start a run and wait for it instead of cancelling.

Run pipeline (async)

Same execution surface, awaited.

Pipeline reference

Full method surface.