@danshipper It's kind of jaw-dropping. I'm in the process of 1-shotting a specialized time/project-management app for myself; within 1 hour (not including time waiting for usage to reset lol) I had a working native app. I ran through it, made a pile of feedback, and it's plowing through it

you just hit the spot.

i think a prompt like this 👇🏼 should be added over each prompt to give more freedom and power to the model.

🤖

" you are an expert at perceiving the present state, identity, intention, ask of the prompter.

observe from a zoomed out lens. call me out. "

@danshipper if you get a reply from fable and not opus you are not being ambitious enough with how you prompt it

if you think fable is incremental you are not being ambitious enough with how you prompt it

@AdarshM88110070 Want to buy it? No one else is paying and I need the $ for tokens

@danshipper Hi today Fable refused to work cause I was swearing. I asked whether it was programmed that way and told me it was his decision. I got chats pic if you wanna see is that normal?

@danshipper hope you get sponsered by them, and stop selling your personality.

@danshipper not really.

if you're 5.5 pro pilled, it's not like fable will knock your socks off.

is it better? in many cases, yes.

if your primary driver is 5.5 pro, though, i'd look at fable as incremental in many cases.

@danshipper > Real. The people complaining about it dont realize how good agentic loops will change the game.

once u see the multiplier effect it clicks fast.

@danshipper Read the room Dan

Ain’t time to be a fan boy

@danshipper sorry but sold personalities are not required by real people.

yeah, 5.5 pro i think is still the best model for many non-technical tasks that aren't centered around single artifacts. if you look at benchmarks like AA-LCR and IFBench, that data shows fable (and all anthropic models) really struggles for tasks that require sustained reasoning across multiple documents and following instructions very closely, which is what a lot of knowledge workers use AI for.

@EverydayAI_ @danshipper I agree with this. I wasn't sure if I was going crazy lol. 5.5 has been extremely reliable for awhile now

@danshipper @mattshumer_

@idbeentaken @danshipper This happened to me with opus 4.7 a few weeks back

@danshipper what i felt was the token burn rate doubling lol

@danshipper have yall published suggestions yet

@danshipper @AdarshM88110070 Haha loved your comeback, Dan. Lots of haters.

@danshipper why do we even need to prompt in the first place??

@danshipper First step is to get fable to build you a local concept-to-prompt expander tool that identifies if your ask or idea is worthy or not worthy and then makes it worthy if it was not worthy

Call it Wayne’s Filter