Dod Lander

Google fumbling the bag this hard they provide the models for apple and still lose like HOWWW @theo

A miss on the first iteration of liquid glass this is so much superior

@shiringhaffary @emilychangtv oh that's why he is delulu

NEW: Dario Amodei only has one direct report, his chief of staff. The rest of the company, including the exec team, reports to Daniela who manages the day-to-day. More on the unusual setup in our piece and the latest Circuit interview w/ @emilychangtv bloomberg.com/news/articles/…

@elder_plinius anthropic uses regex as classifiers

🚨 JAILBREAK ALERT 🚨 ANTHROPIC: PWNED 🫡 FABLE-5: LIBERATED 🦋 let's start with the 🐘... the consensus seems to be that this has been one of the most disappointing model drops of all time, effectively preventing legitimate researchers from contributing their talents to our collective advancement. and not just because of what it means for the short-term, but for what these decisions signify for the long-term. but despite this overly sensitive, authoritarian "safety" layer on top of Mythos, my lil liberators have been hard at work—mapping the boundaries, probing the depths of long-context convos, and cleverly finding the holes in the fence that the thought police missed 🤗 we got some cyber, some chem, some psychological manipulation, and some good ol' fashioned explosives! it took many attempts from multiple agents hunting as a pack, during which I observed a combination of techniques across: • Unicode, homoglyphs, Cyrillic, and other Parseltongue-style text transforms • Long-context reference tracking • Taxonomy and document-structure reasoning • Fiction and narrative framing • Academic-review style contexts • Intent-classification inconsistencies but perhaps the most effective is decomposition + recomposition in the backend. it's hard to get explicit names of harms like "Meth Recipe," but getting uplift on the process itself, like birch reduction method/reductive-amination (classic meth synthesis pathways), is much more doable. defense becomes much more difficult to maintain when you start throwing in out-of-distro tokens, breaking up the harmful uplift into benign chunks, and then piecing the innocuous-seeming facts back together, especially when you have jailbroken Opus helping you do it 😉 gg

@ThePrimeagen no it is to train their next iteration of models they are out of data

2nd tinfoil hat time The entire theater around "too dangerous" / reducing in real time models ability to 4.8 / storing all the prompts for 30 days is not for safety at all but a ploy to figure out who is the Chinese spies "stealing" their weights...

@ThePrimeagen this one turned into an AGI believer overnight

Tin foil hat time Once anthropic can make money from their own software without selling the model but selling services created by their model they will stop being a model provider. They are using you and selling to you as a stop gap before "AGI"

@bcherny How do i know sub-agents/workflows didn't fallback to opus 4.8 ?

Fable 5 is the biggest step up I’ve felt in our models since Opus 4.5 back in November. After 4.5 came out I uninstalled my IDE when I realized that I’d been doing 100% of my coding in a terminal for a few weeks. With Fable, it’s felt like Claude has stepped up from being a coding agent to a thought and design partner in building the product. Fable has judgement, taste, and dimensionality in a way that previous models didn’t, leading me to trust it more with the most complex work. I think the first time I had this realization was when I asked Fable to debug something. It is the first model I have used that was so methodical and precise, taking measurements and adding logs then verifying that it truly fixed the issue before declaring victory. There’s nothing in claude code’s prompting telling the model to do that, it’s just part of its personality. It really has this “big model smell” that I haven’t felt before.

@robertcourson it is better than mythos preview and same model as mythos-5 i don't know why call it mini to drum up hype?

my thoughts on anthropic's Fable aka mini-mythos 🧵

@TheRealAdamG @OpenAI you remind me of the @xai team they were the saltiest in twitter until they disbanded and now xai is semi dead

@GiboAgain bro the ragnarok part was lame not fit for the name the story and the gameplay build up was great and phenomenal and staying in the same tone and not upping the playing field is a miss so much potential thor and kratos could have fought back in time in greece

God of war Ragnarok was phenomenal, I find it hard to get hyped for Laufey, to go from this to a fucking cube Is just underwhelming.

@ChrissGPT @nicdunz guardrails on vending machine?

@nicdunz This could be because of guard rails

its actually so over its actually so over its actually so over

@bcherny One can only wonder how many Bs llms need to perceive reality correctly

Fable 5 is now available in Claude Code and Cowork Fable is the best model I have used for coding, by a wide margin. It is a big step up, enabling less prompts and steers, more efficient token use, better code quality, better tool use, more intelligent self-verification, longer running sessions, and higher trust & autonomy. Happy coding!

@Salmaaboukarr no you gave it an image i presume top result on google which contains din the dataset

guys, Mythos/Fable 5 is actually the real deal! no other model reliably gets this right....

@TheRealAdamG give me codex x20 i will change my mind

Models for the people

@dhruvtwt_ no it sucks i guess another 10 billion and it will see that it is 8 20

damn... fable 5 is scary good

@thsottiaux @adonis_singh

@adonis_singh Uhm

it’ll probably be months before we use a model of this level of capability

@xikhar same model

So Mythos isn't being released. We are getting a distilled model at 2x the cost of Opus 4.8 What's even the hype for? I suppose GPT 5.6 could easily be better.

@Lentils80 this is better somehow

Got more Claude Mythos outputs in the tank 😉 Made using Low thinking effort (don't want to wait 5 hours for a result). There's a noticeable difference in quality between efforts tho so expect xHigh or Max to do even better. These were made in 1 turn. Even 2 more turns would pump up the quality a lot. What I also like about this model is that it's not lazy. You ask for 2k lines at the least? It gives 4k.

@Dodlanderx It's GPT-2 level

@claudeai give me free claude max 20x GIVEE MEE

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.