Sabitlenmiş Tweet
Shikhar
11.6K posts
Shikhar
@xikhar
I talk about tech and AI • Linux enthusiast
Katılım Nisan 2018
706 Takip Edilen3.2K Takipçiler
but "mythos is too scary to release" they said...
Claude@claudeai
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.
English
@notKartikk @cursor_ai It is good. I'm trying not to waste credits on stupid prompts. And if you try to ask even remotely shady stuff, it hits guardrails very quickly and switches to Opus 4.8.
English
Fable 5 is live on Cursor.
I guess it should remain available even after June 22 @cursor_ai?
English
@JeffSte17327059 I'll stick to using gipity for the actual stuff.
Even my company probably wouldn't give Fable now because they want 30 days data retention on all mythos class models for businesses.
English
Cursor is probably a better subscription than Claude.
x.com/i/status/20644…
x.com/i/status/20644…
Shikhar@xikhar
Fable 5 is live on Cursor. I guess it should remain available even after June 22 @cursor_ai?
English
This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward (imo of the same order as Claude 4.5 was in November), peaking especially for long problem-solving sessions on very difficult problems. You can give it a lot more ambitious tasks than what you're used to, the model "gets it" and it will just go, and it's never felt this tempting to stop looking at the code at all (but don't do this in prod!). The model still has quirks that people will run into and the safeguards are configured to be a little too trigger happy for launch, which can hopefully be tuned over time.
I feel a lot of things changing as working software increasingly comes out on a tap. The Jevon's paradox kicks in and I feel my own demand for software growing substantially. You can ask for anything - explainers, visualizers, dashboards, bespoke single-use apps (e.g. a full wandb that is hyper-specific just for your project), you can 10X your test suite, auto-optimize code, run giant research projects with custom HTML for the results, anything! "Free your mind" (Matrix ref). Really looking forward to all the things people build!
Claude@claudeai
Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision. The longer and more complex the task, the larger Fable 5’s lead over our other models.
English
@warkancash They shipped mythos, I was wrong. But it has guardrails, so it's like a better model in general instead of having all the mythos abilites.
English
@xikhar GPT is better for the simple fact that they create mini and nano models that are essential for low-cost micro tasks... If you can't scale usage due to the cost per token then it's useless for average user use..
English
@rahul_twtss They already released it. It's the same model as the latest Mythos 5. But there's heavy guardrails so nothing is getting hacked.
x.com/xikhar/status/…
Shikhar@xikhar
So Claude Mythos 5 is the same model as Claude Fable 5, not a distillation as people were assuming. The catch: - It's not for token-poor people. You have two weeks to enjoy it on subsidy. - For all businesses, they now require a 30-day data retention on Mythos-class models.
English
@xikhar Either there will be memes or stuff getting hacked but it would be fun to see this play out.
English
So Claude Mythos 5 is the same model as Claude Fable 5, not a distillation as people were assuming.
The catch:
- It's not for token-poor people. You have two weeks to enjoy it on subsidy.
- For all businesses, they now require a 30-day data retention on Mythos-class models.
Shikhar@xikhar
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.
English
@ImNotTheWolf China should have the bot centers ready for that already. Next month we'll see mythos distills for a fraction of the price.
English
@xikhar Can't wait for locally hosted models to rip intelligence from this so we can actually have affordable AI.
English
@rileywestreel I was wrong. They have released the new Mythos 5 and Fable 5 as the same model, with more restraints.
x.com/xikhar/status/…
Shikhar@xikhar
So Claude Mythos 5 is the same model as Claude Fable 5, not a distillation as people were assuming. The catch: - It's not for token-poor people. You have two weeks to enjoy it on subsidy. - For all businesses, they now require a 30-day data retention on Mythos-class models.
English
@xikhar wouldn't write it off before benchmarks drop. "distilled" from a Mythos-class model is still a high ceiling
English
@AIMelGibson Yeah, clearly for people who are making a lot of money off using them, and enterprises.
But they have changed their policy to a 30-day data retention for all businesses. I wonder how many will drop support.
English
@xikhar Yeah I knew it was going to be expensive as hell. I mean at this point I just don't think anthropic models are for the people. Much more looking forward to GPT 5.6.
English
@AIMelGibson Damn, they got me this time. Although it will use API pricing after two weeks, so we can safely assume this and any further mythos class models won't be on subscription plans.
English
@Dodlanderx Yeah, it's better than what I had expected. Let's see how it performs now.
English
