Anthropic engineer: "Fable is one of those models you’ll just remember - like Sonnet 3.5, Opus 4, Opus 4.5. We just removed 80% of Claude Code’s system prompt. A new class of models wants a smaller prompt." in 20-minute session, an Anthropic core engineer reveals the capabilities of Fable.... Watch it now, then read how to use this frontier model in the article below.show more

new useful video.thx bro

you are welcome Dipper ! Are you wanting for Fable 5 lauch ?

yep,i speaking with him now

I haven't seen this interview yet, thanks for sharing

This is just from a live session at AI Engineers, bro, less than an hour ago. Brilliant watch on Fable

Fable sounds genuinely game changing.

watched it. worth every minute

worth watching

Yeah bro, a live recording from the AI Engineers panel - definitely worth bookmarking.

smaller prompts for better models is the way

yeah, this is right pipline bro !

Informative!. But he speaks too fast

fable sounds like a reset

@desiAIlabs

Better models often mean simpler instructions. Less prompt. More capability.

Sounds like this presentation context is stolen from a specific users methodology

Great insight! Smaller prompts with stronger reasoning could be a big shift for AI agents. Excited to see what developers build with Fable. By the way, if you like AI and Web3 content, I'd appreciate a follow I'll gladly follow back too!

the harnessshould be dynamically edited at runtime dependent on the task

sounds the right direction similar to a comment i left on here a while ago re interviewing and unknowns. would like to see some commitment to bdd testing to prove the detailed spec you will end up with is implemented correctly. the more detailed the spec the less hallucinations.

Bro, super useful content as always and the view count proves it

The system prompt detail is the real story here — 80% smaller and the model still holds together better. That's not a UI change, that's a different level of reliability under the hood.

removing 80% of the prompt?? wild

the 80% prompt cut is the part id want unpacked, smaller prompt usually means the model stopped needing the guardrails

Nice move. Smaller prompts, bigger thinking. Excited to see how this unlocks faster, tighter models. Congrats on the progress—very cool milestone.

i'm seeing this video for the first time really sick honestly is this some kind of conference??

Smaller prompt makes sense when its job is teaching the model how to reason. Mine's huge and keeps growing. It's not scaffolding, it's memory. Months of decisions and context a fresh model can't reconstruct from scratch no matter how smart it is.

Impressive progress. Thanks for sharing.

The real story isn't Fable. It's that every prompt engineering course from 2024 just got shorter shelf life.

The CC system prompt still needs to be available for other models though?

wild how fast things are moving - was pretty cool to see him show the progression from even just opus 4 -> 4.5 -> fable

Sounds hype - what benchmarks back that claim?

Quality post

fable hitting different, can feel it in the response quality

Why? Looks like it will be a state secret forever. Only the insiders will have it.

$19B valuation, zero revenue. Similar hype cycle?

Impressive vid and just in time after long wait, thanks for sharing 👍

so interesting vid, thanks for sharing

Interesting take. I like the idea of smaller system prompts if it still keeps the model aligned and consistent.