i’m giving openrouter a try now, has most of the models so good for comparing, but token based pricing, curious if that ends up becoming too pricy… although maybe not with a harness like pi! claude code will send 50k tokens in tools for a simple ‘ping’ o.O

self hosted 30B qwen is actually suprisingly capable as well! and glm-4.7 flash (q4_k_m) fits on my gpu and is great (but tends to get into infinite loops :( )

the funny thing is though that there were apparently studies though that show that younger folks especially have an AI-dar

and “that’s so AI” has become a term for kitsch haha

cos the ui is why people abandoned it… :picard:

the audacity (sorry :D)

that time she released that album under the name ‘tool’ and it topped her own in the charts surely must’ve been a career highlight :D

TIL

guess i just never botheres to try cos config change & rebuild wasn’t much more of a hassle :shrugs:

just for clarity: you can modify stuff like hosts or cron jobs but it’d get overwritten iirc? you can also make the change in the config and have it persist (reproducibility being the main point, not disallowing you to edit your files)

meanwhile scavengers reign got cancelled cries

well if it makes you feel better you just made me man history xD

i just cat grep .bash_history lol

but this does sound more convenient xD

welp, looks like you don’t use python virtualenvs… well i guess jokes on you all your shit is probably broken now (and as a bonus, that’s probably a big part of the donwload size as well) :p

electron has entered the room

tbf it probably is still significantly less that windows, i was being a bit facetious…but it’s still at like 1-1,5gb idling on a fresh boot (this is the whole DE, not just cinnamon)…

i did a fresh install with mate on an old machine though and it was a lot less (the usual 500mb or something) can’t see anything suspicious running though - and yes I did check without any stuff running in the background like steam which is also stupidly intensive with their webkit nonsense

does it still use the same amount of ram as windows? :/

deleted by creator

yeh that’ll probably be it tbf… the cuda drivers are specifically for scientific computing and are pretty rubbish for anything else unfortunately… even amd ones are like that :(

however a way i found around it is to just push my gpu compute envs to docker and voila (also avoids the pain of installing the drivers cos nvidia actually provides a cuda docker image) :D