Ah, we specified a minimum version, so I guess that's how it works :) Thanks for reviewing the PR!

There is some satisfaction of having all versions consistent... We could switch to>=X.Ya1 everywhere (taking inspiration from @adamchainz projects, for example)

Not sure about the benefit with >=X.Ya1 but I agree that it looks cleaner when all consistent and my vote is for >=X.Y

Not sure about the benefit with >=X.Ya1

Less churn: we update the matrix once when first adding support and don't bother with following up with other PRs when the final version is out

Yeah, I think it's more of a coding convention than a technical issue, and once you've chosen that convention, I think we'll have fewer problems writing or reviewing code like this in the future. Ideally, the DRF should be released along with the major release of Django, so we could choose the following workflow:

0. (Optional) When a Django alpha or beta is released, add the version to tox testing and leave the pr as a draft.
1. a major version of Django is released.
2. quickly add a full release test to Drf's tox test and merge it into the master branch.

Thanks for reviewing my PR, all! :)