Top
Best
New

Posted by anabranch 6 days ago

Anonymous request-token comparisons from Opus 4.6 and Opus 4.7(tokens.billchambers.me)
615 points | 575 commentspage 8
fny 6 days ago|
I'm going to suggest what's going on here is Hanlon's Razor for models: "Never attribute to malice that which is adequately explained by a model's stupidity."

In my opinion, we've reached some ceiling where more tokens lead only to incremental improvements. A conspiracy seems unlikely given all providers are still competing for customers and a 50% token drives infra costs up dramatically too.

willis936 6 days ago|
Never attribute to incompetence what is sufficiently explained by greed.
rvz 6 days ago||
Correct.
mvkel 6 days ago||
The cope is real with this model. Needing an instruction manual to learn how to prompt it "properly" is a glaring regression.

The whole magic of (pre-nerfed) 4.6 was how it magically seemed to understand what I wanted, regardless of how perfectly I articulated it.

Now, Anth says that needing to explicitly define instructions are as a "feature"?!

alekseyrozh 6 days ago||
Is it just me? I don't feel difference between 4.6 and 4.7
Futurmix 5 days ago||
[flagged]
agentseal 5 days ago||
[dead]
chandureddyvari 6 days ago||
[dead]
EthanFrostHI 6 days ago||
[flagged]
jeremie_strand 6 days ago||
[dead]
contractlens_hn 5 days ago||
[dead]
kziad 6 days ago|
[dead]
More comments...