@ustas_eth@chevonphillip I'd really like a list from OpenAI for "cyber" models (or at least mentions of cyber when announcing new models). It's always a testing game to see if a new model is approved for daybreak blue. As far as I know 6 sol is the "highest", but 5.6 sol is still better imo.
As someone with daybreak blue, i'm so glad that the 500$ plan is coming. I couldn't scale with more accounts since I couldn't get daybreak access on them, so API was the only way (which is not cheap at all) until now.
@ReebootToInit5@Bugcrowd That‘s why I always record a PoC video. That way they can take 10 years for triage and fix the finding, but I can still prove that I was able to reproduce at the time.
PoC videos have probably saved me 7-8times now.
I claimed the "andiqq" handle on @ThePirateFace 🏴☠️
Where Hugging Face AI models never die, and are immortalized as torrents.
Claim yours 👇
https://t.co/ftpUdLgsdY
@0xFireFist It reroutes to 5.6 sol or „gpt-daybreak-latest“, you can check the logs either yourself or with another AI session, you‘ll see that after a long „cyber session with astra“ you will barely have used astra.
@2ap2cb2 If it‘s a ban hammer they‘re losing A LOT of customers. I‘ll just go rogue with DS v4.1 flash, it‘s been doing great for cyber up until now (only tested it some hours but produced results quickly)