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

Claude Fable 5

anthropic.com

Post reply on HN
Print view View original

401–410 of 1001 posts

Re: Claude Fable 5

#401

Post by fbnszb »

knivets wrote: post #314

> Software engineering. During early testing, Stripe reported that Fable 5 compressed months of engineering into days. In a 50-million-line Ruby codebase, the model performed a codebase-wide migration in a day that would otherwise have taken a whole team over two months by hand. How was it measured? How was the output of this magnitude verified over a period of couple of days?

They just went by gut feeling. Classic snake oil marketing haha. No real data to back things up, just let some famous people say they feel better when using it.

Re: Claude Fable 5

#404

Post by OtomotO »

jkelleyrtp wrote: post #33

On the new FrontierCode [1] benchmark (ie graded from an OSS maintainer's perspective of "would I merge this code?") - Opus 4.7 xhigh: 5.2% - Opus 4.8 xhigh: 13.4% - Fable 5 xhigh: 29.3% Seems like a huge jump. [1] https://cognition.ai/blog/frontier-code

Bummer! When can I finally and confidently get slopcode into Zig?

Re: Claude Fable 5

#405

Post by skerit »

bottlepalm wrote: post #377
dannyw wrote: post #255

Impressions from testing Fable 5 prior to launch: • My most noticeable immediate jump was in how its frontend design was much more intentionally crafted, and delightful without feeling like 'AI vibe coded'; with better end-user usability too. • In some internal agentic harnesses, it achieved better results with about half the tokens, making it cost the ~same as Opus 4.8 price-wise! The real price increase is less tha…

I just ran it on a tough reverse engineering problem I'm having that neither Claude Code 4.8 or ChatGPT Codex 5.5 could figure out. 30 minutes later Fable has it all figured out perfectly.

Oh nice, it didn't flag the request? I feared any reverse engineering would become impossible because of the new safeguards.

Re: Claude Fable 5

#407

Post by fabled-out »

Anyone know how to bypass the extremely strict filter Fable 5 seems to have on health/medicine?

I have a rare form of cancer where existing data is very scant/scattered so LLMs have been super helpful to pull together threads across the research landscape. I have an oncologist appointment tomorrow to discuss next steps and am trying to use Fable to figure out some questions to ask my oncologist but keep getting thrown back to Opus 4.8.

My prompt is literally just: My demographics + current treatment plan I'm on including name of my chemo drug + how I'm responding to treatment + "I'm meeting with XYZ tomorrow, what questions should I ask her".

Re: Claude Fable 5

#408

Post by ieie3366 »

UncleOxidant wrote: post #338

> During early testing, Stripe reported that Fable 5 compressed months of engineering into days. In a 50-million-line Ruby codebase, the model performed a codebase-wide migration in a day that would otherwise have taken a whole team over two months by hand. How in blazes do you end up with a 50M line Ruby codebase? WTF?

Very easy. Just have a monorepo and enforce the use of a single language. The company I work in has 1m lines of TS and stripe has 50x our headcount, tracks out pretty well

Re: Claude Fable 5

#409

Post by thepasch »

bkjlblh wrote: post #220

> In light of the ability of recent models to accelerate their own development, we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restrictio…

Yeesh. Anthropic's paranoia about China is starting to get pathological.
Post reply on HN