For running tests in parallel on a single machine, turbo_tests is a great option. There are a few small benefits that ABQ has which I'll detail below.
For running tests in parallel on GitHub Actions, how are you splitting the test files across multiple jobs? Typically, most approaches I've seen pre-partition test files into separate groups. Commonly, that approach results in partitions which are imbalanced, with some jobs finishing before others. Since ABQ is queued-based, it finishes the tests as efficiently as possible. I wrote a little bit more about the difference between pre-partitioning and using a queue under "Reasons for Queue-Based Parallelization" in the announcement blog post:
For the use case of running in parallel locally, one difference between turbo_tests and ABQ is that ABQ wraps the native test framework CLI. So whereas turbo_tests does not support every option that the rspec CLI supports, ABQ does. You invoke ABQ as `abq test -- {any test command goes here}`, such as `abq test -n 2 -- rspec --example 'part of test description'`
ABQ is also a universal test runner, so you can learn to use it once and then apply it across multiple testing frameworks, rather than needing a language-specific solution for each framework that you use. ABQ currently supports RSpec, Jest, Playwright, and pytest.
Dan