franze · 3 days ago

my take: better at findings bugs, worse in conversation and design.

and that finding bugs stuff was exactly the thing where it was neutered...

mwigdahl · 3 days ago

Yes. It is great for analysis, debugging, and understanding larger segments of code. I'll use it sparingly since it won't be on subscription, but there are cases where it will make sense to use it.

bigfatkitten · 3 days ago

I received absolutely nothing worth paying for.

I hit the safety filters with almost every meaningful task I tried to test it against, and these all happened to be benign tasks that Opus 4.8 does reasonably well.

antoineMoPa · 3 days ago

I prototyped some rather complex internal library with it, taking advantage of the free period. It was cool while it lasted, but I wouldn't pay for it. I'd be able to do the same thing with cheaper models and more time + involvement from me. I'd work a bit longer, but would also have more insights about how the project works. It would have cost ~160$ for one day. Hard to justify using it as a daily driver. I did appreciate its concision + willingness to go an extra mile to make sure things work, but all of this could potentially be ported to smaller models with the right training setup I assume.

nnurmanov · 3 days ago

How is it with UI? Any good?

Artexis · 3 days ago

Subscription is my way forward. Fable is an amazing model and reveals its edge over Opus the more I use it. Hard to go back. I feel like they will bring it back to subscription, especially when OpenAI releases GPT 5.6, which is likely this week.

joshstrange · 2 days ago

Nope.

It was awesome to use, I really enjoyed the complicated things I was able to accomplish but in 3 days I burned ~$1,400 in tokens (if I had not been on a subscription) and what I got wasn't worth that. If it was $140 then it might have been worth it.

I'm hoping OpenAI forces their hand and they have to re-include Fable in the subscription to compete.

plkt_studios · 2 days ago

It was burning through the usage limits... sonnet 5 seems very good

LUmBULtERA · 2 days ago

Ping ponging to ChatGPT/Codex for 5.6 launch to test that out... Hopefully its out today or soon.

bescob_ar · 2 days ago

It's good, better than Opus xhigh, but honestly in most cases I can get 'good enough' just with Opus and some iteration. Even Opus 4.7 usually pushed me along toward getting a deliverable out the door. I constantly have 20-30% usage left for all of 5 hour windows and weekly. I'm content to stay on Opus and try codex until Fable comes back.

patates · 2 days ago

It's very cool, but IMHO too expensive for my hobby projects.

It could make sense for large companies to pay for it.

I was already happy with the models since Opus 4.5.

zloy88 · 2 days ago

Nope, not worth it for me. It didn't do any wonders .. it's obviously better, but not how many people act it is. I noticed it especially in big repositories, it does same mistakes as all models - it doesn't check the entire project because it's probably already to expensive to load so much context into a session. So they all struggle the same way when projects getting real big. Anthropic has to work on efficiency from now on, people won't pay so much when every competitor is way cheaper. The first thing you found on the internet once Fable was released, was how to run this model cheaper ... so I doubt the majority of people will pay for it. The only benefit I see in it is paying Fable but use it only for research and let cheaper models implement it.

rstagi · 2 days ago

I think yes, I will keep using it for some analysis and documentation tasks, I think it's been really good at those. Also when I really wanna make sure about the quality of some piece of code, if not too large (otherwise I'll break it down first).

This is only feasible if you have a company and you can spend those money, I wouldn't if it was for personal use