What did you do?

We have a noticed a performance regression on Alerts with the new UI on 3.50 with many firing/inactive alerts (~6k). We never had issues with that with the old UI but with the new UI, it takes more than 5s to load the alert page where it loads instantly on the old one.

For the context, we have several clusters which share the same configuration and on lower-stage clusters, some devices are not present which hence generates many alerts. These alerts are obviously not actionable so they are simply sent to /dev/null.

I have created a reproducer in this repo. There was also a similar issue last year #15551

I have checked with the Developer tools and the new UI makes two identical requests where the old one does only one.

Loading with old UI:

Loading with new UI:

What did you expect to see?

I'd expect the alert page to be as fast on new UI as with the old one.

System information

Running containerized but I don't think that's relevant here.

Prometheus version

prometheus, version 3.5.0 (branch: HEAD, revision: 8be3a9560fbdd18a94dedec4b747c35178177202)
  build user:       root@4451b64cb451
  build date:       20250714-16:15:23
  go version:       go1.24.5
  platform:         linux/amd64
  tags:             netgo,builtinassets