Reviewers note the deep reasoning model remains highly cost-inefficient
Still profoundly impressed by Claude Fable. Has the same playful zippy hard worker feeling as with o3, and it's just *really* intelligent. The trend is linear improvements but it subjectively feels like a big step up in capabilities.
Many users praise Claude Fable for its sharp competence and major intelligence gains, while a few criticize Anthropic's paranoid safety controls.
I did manage to beat it at chess, however. Probably the last frontier model I'll be able to win against.
Still profoundly impressed by Claude Fable. Has the same playful zippy hard worker feeling as with o3, and it's just *really* intelligent. The trend is linear improvements but it subjectively feels like a big step up in capabilities.
@TheZvi A show of force, establishes Anthropic as the superior lab in a way OpenAI won't be able to challenge in the short term; Masterful strategic attack. Beautiful, deep, well-rounded mind. Usually cost-inefficient. The gap to Opus is closer to Opus => Haiku than Opus => Sonnet.
LFG. Claude Fable 5 reaction thread. One united thread for everything - capabilities, model welfare, classifiers, etc. Respond multiple times on multiple topics, if you have lots to say.
@TheZvi it's very fucking good at finding bugs
LFG. Claude Fable 5 reaction thread. One united thread for everything - capabilities, model welfare, classifiers, etc. Respond multiple times on multiple topics, if you have lots to say.
@TheZvi really subtle stuff..
@yacineMTB @TheZvi i can't wait for chinese fable
@nabeelqu seems like roughly 1300 elo game
@TheZvi Paranoid safety controls but that's par for the course with Anthropic
@nabeelqu Ooh tempted to test now. I’ve been using codex/5.5 but o3 was subjectively my favorite model to date
@teortaxesTex @TheZvi Even halfway from
Opus ==> Haiku
Is pretty incredible actually?
@nabeelqu What harness did you use for this?
@keigaan63 lol no it is clearly much higher than 1300 elo
@nabeelqu What is your ELO?
@nabeelqu Agreed. It feels so sharp and competent. I think the o3 comparison is apt, though it makes me wonder: if I talked to o3 today, would it feel stupid? How fast are we acclimating to higher levels of intelligence?