Try failed jobs first

Is there a way to prioritize jobs that failed in previous push?

I have this highly parallel action where many jobs need a single job

(and each of these have a single dependent)

The jobs (of the middle column) get scheduled/queued in the order they appear in the YAML.

I would rather the 2-3 jobs that failed on my previous push get prioritized as this would give me feedback faster.

Note the jobs can be safely queued in any order as they are all siblings.

Is there a way to do this today?

With a bit of work you could write a thing that retrieved the previous results and then ordered your matrix appropriately.