pytest-split

Visit Website

Introduction

pytest-split splits a pytest suite into sub-suites of roughly equal duration based on test execution time. You record durations once with --store-durations, commit the .test_durations file, and then each CI job runs pytest --splits N --group i.

Tests without stored durations are distributed using an average, and algorithms such as duration-based chunks or least-duration balancing are available. It combines well with pytest-xdist: split across machines, then parallelise within each machine.

In the quality loop it reduces wall-clock CI time for large Python suites.

The source code is at jerry-git/pytest-split on GitHub under the MIT licence, with about 330 stars; the most recent commit was on 3 February 2026.

Features

1. Duration-based splits: Balanced groups from recorded timings.
2. Simple CLI: --splits and --group flags.
3. Algorithms: Choice of splitting algorithms.
4. Unknown tests: Average duration for new tests.
5. xdist compatible: Combine with in-machine parallelism.
6. No service: Works offline from a committed file.

Pros

1. No external service: Everything stays in the repo.
2. Free to use: Open source under the MIT licence, so the core tool costs nothing to use.
3. Actively maintained: The repository had commits as recent as 3 February 2026.

Cons

1. Stale timings: Durations file needs periodic refresh.
2. Static split: No dynamic queue like hosted services.
3. Small maintainer team: Releases are occasional.

Usage and Links (Example)

pytest --store-durations
pytest --splits 4 --group 2

Resources:

• pytest-split website
• GitHub repository

Pricing

pytest-split is free and open source under the MIT licence.

Recommended for?

1. Python projects sharding tests across CI jobs.
2. Teams that want splitting without a paid service.

Conclusion

pytest-split is a simple, effective way to shard pytest across CI machines.

Refresh the durations file regularly to keep groups balanced.

On CheckOps, pytest-split is filed under the Release & CI gates stage. Related tools listed here include pytest-xdist, pytest, Knapsack Pro, pytest-rerunfailures, Buildkite Test Engine and Currents. Which one fits best depends mostly on your stack, your team’s size and skills, and whether you prefer to run tools yourself or pay for a managed service. This page reflects the official website, documentation and repository as they stood on 11 November 2026.