GPT 5.1 Instant has trouble completing - UX bug or model issue?

I’ve been trying the new GPT 5.1 option, and I’ve noticed that sometimes when I prompt, the model gets stuck and repeats the last token several times before timing out. Not sure if this is a problem with the model itself or T3’s custom UX. Example screenshot included. I’m not 100% sure, but I think it’s more common on mobile browser than desktop.