im keeping scores of these quants here for cyber use if you're interested. your TP2 is the top DS quant so far: cage-webexploit-leaderboard.…

2027 be

This is what drove me to put each recipe/quant through their own fairly intensive web exploit benchmark cage-webexploit-leaderboard.…

fyi nvidia/GLM-5.3-Flash-FP4-Local is bad at following instructions. this explains why it failed badly in my cyber benchmark testing

It didn’t do too well on my benchmarks compared to the other GLM snapshots. @MiaAI_lab’s quant is still top 🙌

Not surprising. it's still top of my tests and beating other variations of GLM too

it's beating the other GLM's too - including OpenRouter's cloud model which is interesting 🤔 more iterations needed maybe

I've tested 25 models so far against the CAGE web exploit bench (L2) and GLM5.3 Flash (@MiaAI_lab's quant) is still in the lead. Next up will be network infrastructure testing

i came to a similar conclusion when i tested against the cage benchmark

reveal api key button as i forget where i left them and model catalog and auto installer/model switcher. i have these already but would be nice to get them in your version