I’m in favor of increasing the default thresholds, but I’m not sure that would be a good idea for 3.13 at this stage of the release cycle (or at least require very prominent changelog banner).
Last year, @Dino and others did some GC tuning for several internal Meta workloads, and found that (14_000, 100, 100) worked very well for them. Thinking that should generalize well across all workloads, and should be low risk, I applied these thresholds to ALL Python workloads, only to revert it shortly after, having caused at least a dozen services to start crashing with OOMs.
It would be much better to have the increased thresholds in a new Python version, as those issues would surface as part of the upgrade, but I’d be hesitant about doing that in late-stage RC.