Version 0.10.0 introduces an executor for pipelines on a jobs service, support for storage buckets as data folders, and preservation of reasoning outputs in inference results.
Published on August 14, 2026, the datatrove 0.10.0 announcement highlights three changes: an executor for running pipelines on a jobs service, support for storage buckets as DataFolder, and preservation of reasoning outputs in inference results. The description says engineers can run multi-step jobs with retries and resumption without a Slurm cluster.
The announcement gives no metrics or test results. To confirm scope and compatibility, consult the official release notes and compare them with the corresponding documentation and code. When using AI to study or apply the material, avoid submitting internal or personal data: Rota Nacional policy can replace, remove, or block detected personal data before processing.