The key part of the error is:

Failed to CreateArtifact: Artifact storage quota has been hit.
Usage is recalculated every 6-12 hours.

If you've already deleted all workflow artifacts, then one of the following is likely happening:

Storage hasn't actually been recalculated yet. Although the message says 6–12 hours, in practice it can occasionally take longer before the quota is updated.
Artifacts still exist somewhere. Your cleanup script may only have removed artifacts from repositories you own. Check for:
organization repositories
forks
repositories where you have Actions enabled but aren't the owner
expired artifacts that haven't been fully garbage-collected yet
The quota belongs to a different billing scope. If the workflow is running in an organization repository, the organization's Actions storage quota is used—not your personal account's.

To troubleshoot, I'd recommend:

Verify the current Actions storage usage in Settings → Billing & plans → Actions (or the organization's billing page if applicable).
List artifacts with gh api (or the REST API) for every repository involved and confirm that none remain.
If the billing page shows 0 artifacts but uploads still fail after waiting more than 24 hours, this is likely a backend accounting issue rather than a workflow problem.

At that point I'd open a GitHub Support ticket with:

the repository name,
the workflow run URL,
the timestamp of the failed upload,
and the exact Failed to CreateArtifact message.

The Node 20 deprecation warning and the punycode deprecation warning are unrelated—they don't prevent artifact uploads. The quota error is the actual failure.

Sometimes the upload step is only where the workflow stops, while the actual issue happens earlier. I'd verify that the files are being created with the expected names and in the expected location before the artifact step starts.

Another thing worth checking is whether the path is being built from an environment variable or matrix value. A small typo or an unexpected value can make the action look in a directory that doesn't exist.

If you can share the workflow snippet around the build and upload steps, it will be much easier to spot what's going wrong.

Based on the screenshot, the error makes more sense now. GitHub still believes your Actions storage is completely full (0.5 GB / 0.5 GB), even though you've already deleted the artifacts.

A few things I'd check:

1. Click View details on the Actions billing page and verify whether the storage is actually attributed to ffmpeg-kit-next (or another repository). Since almost all of the usage comes from that repository, it may still have artifacts or caches that haven't been fully reclaimed.

2. Verify through the API as well, not just the web UI. Sometimes the UI shows no artifacts while it's worth confirming with:

 gh api repos/OWNER/REPO/actions/artifacts

and, if applicable:

gh cache list

to make sure there are no remaining caches.

If both the UI and the API show nothing, but the billing page still reports 0.5 GB after waiting a week, this doesn't look like a workflow configuration issue anymore. It looks more like the storage accounting on GitHub's side hasn't been updated correctly.

In that case, I'd recommend opening a GitHub Support ticket and including:

the billing screenshot showing 0.5 GB / 0.5 GB usage,
the repository name (ffmpeg-kit-next),
the failed workflow run URL,
and the output of gh api repos/OWNER/REPO/actions/artifacts.

That should give the support team enough information to verify whether the storage quota is stuck and reset it if necessary.