T3 Stack creator Theo Browne defends Claude against recent criticism, calling performance complaints unfounded

This post is just blatantly false. Theo isn't anthropic's #1 hater, I am.

@theo Theo is unable to lie ( he’s on the spectrum) . So trust his opinions.

@richa_lq Not wrong

@theo It’s a good model

@theo correct

Is it just me, or is it really bad at coding on an existing codebase? It seems to get stuck in the patterns that are already there and can't be creative. I compared it to much weaker models and it didn't do as well on the same prompt (Make a page in a similar style to X for a new audience.)

But when doing kind of like open-ended blue sky research, it's the best type of research I've ever seen. I've never seen something so comprehensive and accurate (tested it in fields I have deep expertise in)

Fable is extremely good and honestly there’s probably extreme demand for it so I even get the rate limits and whatnot

Maybe when GPT 5.6 Sol hits some usage will move away from Fable/Mythos and they’ll be a little more generous

But at this point I bet these consumer plans are like 10-15% of their actual usage

@theo It’s cause of this guy. Who chronically posts bad benchmarks and takes,

@theo Agreed. There are plenty of valid reasons to dislike Anthropic, but Fable is definitely not one of them. However, running the model on "high" mode and not any higher is the way to go.

@theo Sam Altman holding you at gunpoint at the OpenAI booth "Yea write that tweet down boah"

@theo That’s what worries me about 5.6 sol. However intelligent it may be, I don’t think it will be as creative as Fable. Being smart is one thing, but being creative & smart is how you solve problems, and Fable has solved problems I’ve spent months working through with ChatGPT models.

@theo The worst thing you can do about something worth legitimate criticism is to start making up shit about it

@yacineMTB nah he makes the videos and stuff. you just yap on here

@theo What wrong thing does Anthropic do exactly?

@theo its a good model sir

@kenwheeler Because it's not optimizing to be loved, it's optimizing to not be hated (much like golf)

@theo The funniest part is that bad criticism protects them. If people keep pretending the model is trash, every serious complaint about Anthropic gets easier to dismiss as cope. 👇

@theo all they doing is spamming with worksloped dumb three.js games and mediapipe crap. They look like chimpanzees being able to do fire by first time in their lives.

@Dodlanderx @richa_lq @theo It could be autism; autistic individuals are often deeply loyal to their core values and won't compromise them. Both of my children are like this.

@theo im just curious how many people will continue to use it after they actually have to start paying for it at cost. may the subsidized subscriptions last forever and ever and ever and ever 💞