>this kills the software 'engineer'good riddance
>>108465156But can it get through Victory Road?
>>108465164no. that's the model after thte one that cures cancer.
>>108465156usecase?
>>108465156>release model with new name>"this kills software engineer">is expensive slot machine for code>waste 6 gazillions monies>throttle and lobotomize it>it's now worse than previous models>slash usage to a quarter>gaslight users, "skill issue">profit? (you wish)>repeatyou niggers will live this forever
>>108465156I bet its slightly better and they just nerf opus if I had to bet.
>>108465156Based as fuck. Codetrannies BTFO.Now I want to see them dying and suffering
>>108465156LUDDITES LOST
>>108465708>>108465709Why are you misanthropic?
>>108465897You bullied me when I was a kid, but my PC didn't, it treated me well. Came puberty, you didn't have sex with me, but my PC did. Came adulthood, you didn't give me a job, my PC did. So fuck you.
It's still all just marketing fluff
>>108465709>hurr durr ludditeIf you use this phrase in the context of AI you should be castrated.
Thanks but I'm still not foregoing personal skills in favor of buying your cloud service subscription
please please please VCs, please, we can pay it back... WAIT NO PLEASE WE HAVE.... MYTHOS! PLEASE DONT SELL
>>108465203The ai cancer-curing model won't be an LLM, it'll be something like AlphaFold.
>>108466124these niggers can't even "sell"not until anthropic IPOs
>>108466083this but you luddites instead
Total Claude victory.
>"Hello!">You have hit your usage limit - resets in 2 months
>>1084660831 "ludditie" for every 3 "bubble" is fair
>>108465928>some random corporate AI = my beloved PCPoor summer child
>>108465156>by far the most powerful AI model we’ve ever developedTHEY DID IT AGAIN
>>108465156This pattern of AI companies describing their own products as so spectacularly effective that they're dangerous really is a remarkable piece of propaganda engineering. What is happening here would be easily understood and obvious by everybody if it the head of marketing for a food company was on TV talking about how pretty soon everybody will be eating their food, and how it's so unbelievably tasty that it might cause people to leave their families and abandon all other hobbies in pursuit of their delicious product, utterly destroying society as we know it.I mean maybe it'll happen eventually. Maybe we'll all end up with a wire stuck in the back of our heads, floating in a vat of nutritious goo. But what we've seen so far has been an excellent, highly useful, and certainly groundbreaking industrial automation product.Maybe they could just write a blog post telling us what the thing does and how much it costs and when we can try it.
>>108465164In the deepest darkest "it's so fucking over," we found the brightest "we're so fucking back."
>>108465156buy an ad
>>108465156>ai getting so good it comes with a warning label nowcool
Claude Code is so much better than OpenAI Codex, but right now chatgpt 5.4 is just more competent than opus 4.6. I'm looking forward to this release so anthropic can catch up and I can ditch the garbage that is codex
>>108465156any metrics on how good capybara is?
>>108470635It's finetuned to be better at TTC. It'll never be worth the price unless you're rich, maybe you can pay $50 for one prompt if you're really stuck on something. It's literally just Claude 5 Opus but more expensive. Just wait a year.
>JUST TWO MORE WEEKS
>More expensive than Opus>The model that costs them $5,000 for every $200 sub
sorry, but as a UI/UX engineer these things just shit out the worst looking UI i have ever seen, and i don’t expect this to be much better. i’ll check back in two years, hopefully by then it’s not still styling navigation links as cards. but you have fun anon. i get that nocoders are mind-blown by AI
>>108470928Imagine being a frontend webshitter and being snobbish about itBro a 16 year old Indian can do your job
>>108466192What is it with tpot twittershitters and the claude mascot making a :3 face?
>>108465156Oh great, now they're targeting Cybersec jobs
>>108471532It's just a cute mascot.
>>108468151mythos is gonna be extremely slow and expensive, like that o3 pro or whatever oai had, they will shut it down in few months and release safe mythos distil that's pretty much exactly like opus 4.6 just costs 2x instead of 50x
>>108471822I don't really care, my company pays for the $200/month plan. On codex I only use xhigh reasoning for everything because I can (my company pays for both obviously).Now if it's so expensive that they literally don't even make it available in Claude Code, then yeah rip I guess.
>>108465164mythos 1.5 will, look at how fast it got to the third puzzle this time bros! JUST ONE MORE TRY!
>>108471822Nope, it will cost way more than 4.6 even lobotomized, we've seen this before.
>>108465156>new model that's supposedly leet at hacking>cybersecurity stonks go downI don't get it. Do people just give up or is this another US White House market manipulation?
>>108465709>>108465928This.Luddites are seething and coping. They already lost, some of them just don't realize it yet.
>>108465156They are just going to rename opus to mythos and charge 10x more
>>108472365crowdstrike has 10k employees.how many do you think they'll need when claude is better than most of their technical staff?
>>108472491If anything fewer employees means more profit. I don't see why the stocks would go down.
>>108472506the more you can automate, the cheaper the works becomes and revenue comes downa lot of cybersec work just gets done in-house because you can just ask claude to do itestablished firms suddenly face competition from a range of startups with 3 dudes offering the same services for a fraction of the cost and the market becomes diffuseand there's the long term view:after an initial period of very heavy exploitation discovery and patching (which will also act as great training data for the RL pipeline), software in general is more hardened and new software agents produce in the future is far more secure by default
>>108465156that entire thing was written with AI. can't stomach itI'll test it when I get back to work
>>108472547Cybersecurity companies survive and grow on street creds. They ask you for your own published CVEs and bug bounties in interviews at big shot companies. Contracts are also signed based on this, like how many CVEs were published under your company? But you do have a point, SMEs not in the know may not care about that and just go with whatever cheapest.
>>108472567test what? it's a leak of an announcement of a limited pre-release, we're in that moment of the hypecycle
>>108465897how can i be misanthropic when i love anthropic and everything theyve done
>That's why our release plan for Mythos focuses on cyber defendersaww, I was getting a lot of well paying short term work thanks to AI induced vulnerabilities popping up everywhere from lazy vibecodingpooey
>>108471868Pro models take half an hour per question on the $200 OAI plan.
>>108472922tb h if they're right the models are going to outpace any human / firm at discoving vulns so i don't know if that metric is going to mean much in a few yearshttps://youtu.be/1sd26pWhfmg?t=533
>>108473114Is the api really that much faster?
>>108473147I don't know, the Pro API takes like $100 per request so I'm not finding out any time soon.But I doubt API is much faster than web.
>>108465156>dumb down current model>new model seems much smarter than the old one despite being marginally better >BREAKTHROUGH IN AI
>>108465156>but!, we must lobotomize it first!fuck these niggers
>>108465164
>>108473114I dunno, works for me. Not sure what the "pro" model is now, the best I can select in the codex TUI is 5.4-xhigh, do they have better stuff that's not available at all from codex?It does take very noticeably longer than claude 4.6 even for simple questions, but this literally does not matter in daily usage, at least the way I use it. It's not half an hour per question btw, it's usually 1-3 minutes to get back an answer, maybe half an hour for it to go around and implement some prompted feature going through multiple steps of reading files, code generation, running the compiler/tests and re-reading the files to review itself. (On the $200 plan of course.) The codex sandbox, system prompts and literally everything else being utter dogshit matters way more.
>>108466693https://news.ycombinator.com/item?id=47542068
>>108473090don't worry anon, most companies won't get mythos licenses for their pajeets if it's significantly more expensive than their current licenses
>>108472506>Claude, what does crowdstrike even do anyways?>[blahblahblah]>Can you do that for my organization?>Sure thing boss.
>>108472922Claude will have a higher CVE count than would be humanly possible. Uncs relying on reputation will get their faces ripped off.These hypothetical cheap startups with 3 dudes running claude that the other anon mentioned won’t even exist either. They’ll be superfluous and a trust risk.It will only be Claude.
