Skip to main content

Create Cluster Run

POST 
/datasets/:dataset_id/versions/:version_id/cluster-runs

Start a clustering run for this dataset version, or return the one already running.

Commits a queued job and returns immediately with the pending run; the actual clustering happens asynchronously through the standard embedding pipeline. If a run for this version is already pending, claimed, or running with the same parameters, that run is returned instead of starting a duplicate. A run already active with different parameters is rejected — wait for it to finish before submitting another with new settings. At most one clustering run may be active per workspace at a time, so a run active on another version of the same workspace also blocks a new one here.

Request​

Responses​

Successful Response