pull down to refresh

70 sats \ 9 replies \ @SimpleStacker 30 Jul \ parent \ on: Stacker Saloon

Indeed. And it tries to sound obnoxiously smart and confident while lying

write
compose
1 sat \ 8 replies \ @sox 30 Jul

That. I just stopped using Opus once Fable came out, I don’t think it ever lied to me.
Also Sonnet feels more reliable than Opus.

reply
126 sats \ 7 replies \ @SimpleStacker 31 Jul

Opus definitely regressed since 4.6, which was GOATED

Haven't tried Fable. Will try. And yes, Sonnet 5 feels more reliable than Opus 5 right now.

reply
1 sat \ 6 replies \ @optimism 31 Jul

Are you using xhigh with Opus? Anything under xhigh is nerf'd. Fable is fully reduced on all paths.

The thing they did between 4.7 and 4.8 was change the meaning of "effort".

reply
147 sats \ 5 replies \ @sox 31 Jul

Opus xhigh still lies, and in general Opus is always so protective of its lies, idk.

I used to love Opus, likewise up until 4.6

reply
122 sats \ 4 replies \ @optimism 31 Jul

Fable lies too, GPT lies too.... GLM also and Kimi is 2.3T lies.

"Protective" means you argue. Don't argue.

reply
147 sats \ 3 replies \ @sox 31 Jul

everybody lies, but I mean Opus really lies all the time

reply
104 sats \ 2 replies \ @optimism 31 Jul

To what prompt?

reply
126 sats \ 1 reply \ @sox 31 Jul

I truly don't know, it seems like that tasks that require more than 10 minutes are more prone to this behavior. Changing the effort level helps a little but not so much, it's like it gets convinced about a fact during reasoning and never checks it again.

The weirdest part is that fighting Opus on presented lies becomes a gaslighting game and it's really frustrating lol. Fable is much, much slower but it's less prone to lies (never happened to me thankfully) and absolutely won't fight me if I say that 1+1=2 (simplification)

view all 1 replies