Swaymak.AI

@darrenhill @JoshKale You have no clue. But it’s fine. You will find out soon enough.

@SwaymaKdotAI @JoshKale Where did I say AI isn’t a good tool? It’s just not AGI. I’m sorry to burst your bubble.

This is big... Anthropic just announced a model so powerful they won't release it to the public out of fear over the damage it will cause 😨 Claude Mythos Preview found thousands of zero-day exploits in every major operating system and web browser... The numbers are hard to believe: > $50 to find a 27-year-old bug in OpenBSD, one of the most security-hardened operating systems ever built > Under $1,000 to find AND build a fully working remote code execution exploit on FreeBSD that grants unauthenticated root access from anywhere on the internet > Under $2,000 to chain together multiple Linux kernel vulnerabilities into a complete privilege escalation exploit For context: these are the kinds of findings that previously required elite security researchers working for weeks. Anthropic engineers with no formal security training asked Mythos to find exploits overnight. They woke up to working code the next morning. The results were so impressive Anthropic assembled Apple, Google, Microsoft, Amazon, NVIDIA, and seven other organizations into Project Glasswing: A $100M defensive coalition. They're not releasing this model publicly. Instead, they're racing to patch the world's infrastructure before models like this proliferate.

@darrenhill @JoshKale You sound like the horse seller when the model T came out.

@SwaymaKdotAI @JoshKale I do this for a living. It’s not AGI.

@darrenhill @JoshKale Keep thinking that. Nothing like sticking your head in the sand.

@SwaymaKdotAI @JoshKale Know any decent AI plumbers or electricians? Must be available for a couple of days next week. Have a few jobs that need doing in my kitchen. (Buddy, we’re way off…)

The Mythos system card was a mind blowing read. If you want to know why they aren't releasing it just yet take a look at just some of crazy things Mythos was able to do: • Found and exploited zero-day vulnerabilities in every major OS and every major browser. Autonomously. No human guidance after the initial prompt. • Found a 27-year-old bug in OpenBSD, an OS famous for being secure. Cost: under $50 for that specific run. • Found a 16-year-old bug in FFmpeg that every fuzzer and human auditor missed since 2010. • Fully autonomous remote code execution on FreeBSD: found the bug, wrote a 20-gadget ROP chain split across 6 packets, got root. A task that would take expert pen testers weeks. • Chained 3-4 vulnerabilities together on Linux kernel for privilege escalation. Bypassed KASLR, heap spray, the works. • Browser exploits: JIT heap spray → sandbox escape → kernel write. One webpage = full system compromise. • Found bugs in TLS, AES-GCM, and SSH implementations in major crypto libraries. • Opus 4.6 had near-0% exploit success rate. Mythos: 181 working Firefox exploits vs Opus's 2. • Thousands of unpatched critical vulns discovered and being disclosed. 89% of severity assessments matched by human experts. • Reverse engineers closed-source binaries and finds vulns in them too. 2026 is about to get wild!

@darrenhill @JoshKale ai is already driving cars much more safely than humans. If you are talking about making novel scientific discoveries... well just wait it's about to get crazy.

@SwaymaKdotAI @JoshKale Can it drive your car? Can it solve the theory of everything? Then it’s not AGI. Sorry I had to spell it out for you.

@EmEyeTea @JoshKale Not at all. Just funny someone who is all a lot worthless fake money is anti ai. Makes a lot of sense if you really think about it.

@SwaymaKdotAI @JoshKale Is that supposed to be a burn or something.

@darrenhill @JoshKale Well thought out logical argument.

@SwaymaKdotAI @JoshKale Erm no

@EmEyeTea @JoshKale Says the guy believing in crypto 🤣

@SwaymaKdotAI @JoshKale Rubbish.

@bubbleboi Agreed! It’s a good model.

Just got access to Claude Mythos… & ughhhhhhhhh this is AGI. It was the first time a model one shotted a 10/25G Ethernet MAC/PCS, it even knew to select the right line rate and data width for lower latency. This alone is something that would take a really skilled digital designer 3-6 months if they had experience in the past to pull off… But it didn’t just do that I then said to make the MAC fully cut through and only forward certain IP addresses within a range downstream it one shotted it instantly also which blew me away… Then finally I thought ok let me trip it up so I said now do 50G MAC and it knew without me telling it to add another GT transceiver and it even added alignment markers and FEC to it correctly. 💀💀💀 It’s passing all the tests I have so I’m going to flash the board and see if it actually works on hardware now…

@immaculatehole @JoshKale 🙄

@SwaymaKdotAI @JoshKale learn the definition of AGI lmao. far from it

@ashiiirg @JoshKale Agentic coding and operations mostly. The contextual understanding shows up when you give it complex multi-step tasks and it holds the full picture without losing the thread. Hard to share specifics but it just doesn't drop context the way other models do.

@SwaymaKdotAI @JoshKale What modality was your primary use case for testing? Any safe to share examples of the contextual understanding?

@egregious_angst @JoshKale My “day job” gets me access. The reasoning leap is what makes voice sound more intelligent and less robotic.

@SwaymaKdotAI @JoshKale They gave you access? Also voice? Like advanced voice mode but better?

@rand_longevity 🎯

forget AGI, we are talking about ASI now

@JoshKale Disagree. We've been blown away by it. Doesn't make splashy images but the reasoning, voice, and contextual understanding are on another level. 83% CyberGym vs 66% for Opus 4.6. Calling this "just good at code" undersells it massively.

@SwaymaKdotAI This model is excellent at code and much less so other modalities

This is AGI by any reasonable definition. Takeoff has started. 2026 and 2027 are going to be unrecognizable.

@DarioAmodei It’s a good model! Well done.

I’m proud that so many of the world’s leading companies have joined us for Project Glasswing to confront the cyber threat posed by increasingly capable AI systems head-on. x.com/AnthropicAI/st…

@chatgpt21 Yeah it’s good but so expensive. 😭

OMG MYTHOS JUST DROPPED A NUKE ON OPEN SOURCE 😭

@alexalbert__ It’s a good model!

We released Claude Opus 4.6 just two months ago. Today we're sharing some info on our new model, Claude Mythos Preview.

@iruletheworldmo People are going to love the model it’s a good model. People are going to hate the pricing! 🤷‍♂️

huge huge day today. times are changing.