[sd9]

Weekends are the untapped frontier. Still room to scale.

[ahstilde]

yup! When i did an analysis last month, GitHub is up 89.3% on weekdays and 96.5% on weekends. Incidents touch 62% of weekdays and 11% of weekends. Claude shows the same pattern: 92.5% weekday, 97.8% weekend. Tuesday through Thursday is the danger zone. Sunday is practically a different service.

https://www.aakash.io/tech-chase/github-and-claude-are-down-...

[jrumbut]

I had an occasion recently where I was working a lot of late nights/early mornings with AI use. And I'd be getting these instant, beautiful responses, and then, as soon as the sun started coming in the windows, it would take longer and fail more, and by the time the clock struck 9 AM, every LLM had turned back into a pumpkin.

[user_7832]

Which service(s) were you using, if you don't mind sharing?

I'm curious if most of the big players including eg Google do this thing of nerfing models or it's limited to more "smart" (read: black box models like ChatGPT.

[jannes]

Are you saying US data centers idle in the night rather than serving European/Asian users?

[zemo]

ideally european/asian users would hit european/asian servers, so potentially not surprising

[pojzon]

Inference results for Copilot are also a lot better during weekends than workdays. Its my personal experience so take it with a grain of salt, but I work on personal projects only on weekends mostly due to that brain drain mon-fri of copilot.

[skor]

change is the biggest cause then?

[sifex]

Or usage

[atrettel]

I imagine that it could be usage, but it also could be fewer people caring to report issues on the weekends too for that matter.

[kshahkshah]

Wait until they go 996!