Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Fable 5 is Back

twitter.com

Post reply on HN
Print view View original

111–120 of 438 posts

Re: Fable 5 is Back

#111

Post by matltc »

stavarotti wrote: post #80

I'll be using it tonight but grudgingly so. Grudgingly because after July 7th, I'm not going to all of a sudden, start paying API prices (and maybe that's the problem) when I'm used to a subscription that gives me multiples in comparative value. Perhaps this is the fabled "token economics will come for everyone this year" that I've been reading about? In any case, I'll use the hell out of it to extract as much as I c…

I locked my default model to opus 4.6 around the time of the nerfs. Such better results compared to 4.7+

That's enshittification for ya I guess

Re: Fable 5 is Back

#113

Post by hirvi74 »

Congeec wrote: post #38
petra wrote: post #26

Maybe, for some projects, instead of generating code with it, it would be useful to generate a plan and the loop(tests/formal verification),because those take much less tokens than a full project, and than use the loop using the older models ?

Yes, I've been using Opus to write a plan and fanout sonnet subagents to implement it. Cheaper and faster

What about quality? Being cheaper and faster, while great and all, is less valuable than quality to me.

Re: Fable 5 is Back

#114

Post by holoduke »

echelon wrote: post #64
aroman wrote: post #41

Earlier quoted context omitted.

This makes me think they really are quite capacity constrained at the moment. I had assumed they were primarily limiting it to entice people to upgrade, but I feel like these limits are so low and so temporary (especially over July 4th weekend in the US) that people will barely get a chance to get "used to it" and then think: "man, I can't live without this, I'll pay for API pricing".

If it's API pricing, I'm going to ditch Claude Code and switch to a harness that can jump between GLM and Claude Code. Cheap pricing is why I use Claude Code. The minute they fumble that, I'm using Chinese models for 90% of the work.

I dont believe they can afford to switch to API pricing. Everyone will leave. I am easily spending the equivalent of 1000 dollars a day on tokens with two max subscriptions. that about 400 dollars a month. Thats acceptable for my position. But thats like 30k per month. Totally not viable.

Re: Fable 5 is Back

#116

Post by iririririr »

meco wrote: post #77
petra wrote: post #26

Maybe, for some projects, instead of generating code with it, it would be useful to generate a plan and the loop(tests/formal verification),because those take much less tokens than a full project, and than use the loop using the older models ?

This is the goal behind Devin Fusion, pretty good results so far I think. https://cognition.com/blog/devin-fusion

so, pretty much undo the "magic" that the harness is for

Re: Fable 5 is Back

#117

Post by iririririr »

LoganDark wrote: post #109
anotheraccount9 wrote: post #103

Not exactly. Every time I ask something using Fable, it switches to Opus. The subjects (and I tried many) seem to be irrelevant. Basically, I can't use Fable.

Check your CLAUDE.md, etc. for any of the banned topics or anything adjacent. Also check your recent Git commit history, if you're running the agent in a repository, as well. The classifier trips on anything anywhere in context, so those could be sources of rejections.

adding rules for the agent to ignore version control and never touch git files or commands (except for git log) improve my slop coding 100%. no risk of sneak commits and much less token wasted.

Re: Fable 5 is Back

#118

Post by holoduke »

mlitwiniuk wrote: post #10

I have to admit that when it was blocked, I canceled my max plan and asked for a refund. It felt like someone took away my previous toy. So I'm happy it's back again; I upgraded to max again. Coding aside, but Claude Design is phenomenal - for both new designs and redesigning existing UIs. So my customers will face a new wave of refreshed screens all over the place in coming days ;)

Yep claude design is nice. You can also do it from CLI or your own lovable clone. I made a fork from claudable https://github.com/holoduke/Claudable What I find stong with Fable is that it can pixel perfect copy existing sites or designs. And it can do complete conversions from plain html to nuxt, angular or react apps.

Re: Fable 5 is Back

#119

Post by vonneumannstan »

trunnell wrote: post #112

Their post detailing the timeline and their actions since reinforce my belief that Anthropic is among the most trustworthy AI companies. https://www.anthropic.com/news/redeploying-fable-5

Its like saying Pinochet was one of the least murderous authoritarian dictators. The bar is low...

Re: Fable 5 is Back

#120

Post by LoganDark »

iririririr wrote: post #117
LoganDark wrote: post #109

Earlier quoted context omitted.

Check your CLAUDE.md, etc. for any of the banned topics or anything adjacent. Also check your recent Git commit history, if you're running the agent in a repository, as well. The classifier trips on anything anywhere in context, so those could be sources of rejections.

adding rules for the agent to ignore version control and never touch git files or commands (except for git log) improve my slop coding 100%. no risk of sneak commits and much less token wasted.

Claude Code automatically adds some of the most recent git commit messages to the system prompt. This was discovered when having git commit messages containing any OpenClaw information would result in Anthropic billing your extra usage (since they have a vendetta or something)
Post reply on HN