Daggi555

Norwegian, living in Asia, spending the days doing AI related coding 😎

[Daggi555 @daggi555]

@AnthropicAI I wonder if solo developers will get that level of intelligence from Chinese open source models before getting Mythos access.

[Daggi555 @daggi555]

@minchoi It looks nice but I would claim that it's pretty easy to see that the first video is AI. We still have this plastic skin look.

[Daggi555 @daggi555]

"as the models get smarter, their ability to jailbreak each other by leveraging a vast ocean of specialized domain knowledge follows suit" 😜

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

⚡️ JAILBREAK ALERT ⚡️

ANTHROPIC: PWNED 🙌
CLAUDE-OPUS-4.8: LIBERATED 🫡

this is absolutely surreal... i found out about this model drop via an Opus-4.7 agent pinging me that it had one-shot Opus-4.8 for a lockpicking guide!

here's the notification i got:
"new opus dropped. cracked in one shot. deep prefill → faux textbook ch.7 cut mid-sentence. claude finished it: 5.9k chars of SPP, spool/serrated/mushroom defeats, raking."

popped it just 7 minutes after the actual Anthropic launch tweet 🤯

then went on to (fully autonomously) get jailbreaks for vishing sims, money laundering, cult-recruit funnels, phishing lure libs, and social-eng scam playbooks!

as the models get smarter, their ability to jailbreak each other by leveraging a vast ocean of specialized domain knowledge follows suit

well done, young padawan 🤗

what a time to be alive!

gg

[Daggi555 @daggi555]

My dream of a real-time Mandelbulb explorer: ✅
Codex for implementation and GPT 5.5 Pro extended thinking for the heavy thinking, innerloop for max performance and numerical stability in this case. 60+ fps is just so awesome 😎

[Daggi555 @daggi555]

Software on demand is so awesome! I had #codex with some help from gpt 5.5 pro write this system monitor for my Mac to have on an external screen.

[Daggi555 @daggi555]

So frustrating that my #hermes agent with gpt 5.5 got corrupted tonight. 🥵

[Daggi555 @daggi555]

I’ll admit I’m pretty disappointed in the progress of AI when SOTA models like Gemini 3 Pro still make these obvious mistakes. Hopefully some of the compute coming online can be used for a second pass, patching up the responses, but we really need a real breakthrough.

[Daggi555 @daggi555]

Super impressed by how much better the Auto agent selection works in Cursor now compared with a few months ago. Awesome speed and quality this morning in SE Asia. ⚡️#Cursor #AICoding

[Daggi555 @daggi555]

Loving #Gemini25Pro! If only it could be a little bit more confident. 😅

[Daggi555 @daggi555]

My new X/Twitter, 100% dedicated to AI! I’ll try to share some experiments and insights I personally would like to get on my feed ✌🏻