With 5.1, there practically seemed to be no difference in ability between opus and fable, but I hope there is a difference this time. When it seems to struggle while working I end up using astra, so I hope it works well.
i've been trying to use sonnet and opus to keep my usage down on the low max plan, but i noticed today that somehow i'm 80% through my weeks usage. so i'm hesistant to even try fable 5.2 yet, but finding the right set point of usage vs model size has been a bit struggle for me, one of my biggest issues has been trying to pull back doing exhaustive testing or other such things that chew through context, and instead have programs that just measure things are working, but i continously seem to run into issues with claude wanting to repeat things for itself in an adhoc manner. i've tried fable directing opus or sonnet subagents, and the reverse. and i found that part of the issue with smaller models trying to direct subagents is that they over specified what they thought the problem was rather than the problem. like they'll have an "idea" about it which is wrong, and then try and feed that into the subagent.
The biggest relief is that the price has dropped by 20%. It’s still expensive, but with my $100 plan, Fable’s usage would spike instantly the moment I started using it, so it was always a source of fear.
I will still use opus for general coding design, but for major designs.
My hypothesis is the opposite, that they don't have anything better.
Look at OpenAI, they've been outclassed by Opus 5.5 two months ago; obviously they don't yet have a response.
Now Anthropic is releasing this lackluster Fable 5.2, a week or two before their IPO.
It's barely an improvement at all, makes exactly the same mistakes as Opus 5.5. Not even worth using with the speed and cost penalty.
The much rumored internal Anthropic model is listed on the Fable 5.2 system card as an even smaller improvement over Mythos 5.2 than Fable is over Opus 5.5.
I expected a larger version of Opus 5.5 would be a significant improvement to launch before the IPO. Turns out it didn't scale and the improvement is marginal.
Note that Claude models could still do better in some cases, just that because of their refusal to answer sometimes and some weird formatting/stupid errors they make sometimes, they do fail more tests than they should.
In practice they might recover from those errors themselves if they notice it, so they would have better results, but the tests are mostly one-shot and they can either get it wrong or right, no 2nd tries (apart from the agentic ones and a few coding ones which have a limited number of turns available too).
That being said, the pricing and speed are accurate, regardless of the test results.
I checked which tests it failed, and it failed some very basic coding question (to mention which line a but is on, got the wrong line, and it did so in multiple such tests and accross different retries, weird).
I agree, subjectively. I am still satisfied enough with Opus 5.5 I find it hard to change. Typically I would find myself willing to try out the latest & greatest so-called SOTA, but I kind of have "enough." I could get more out of Opus 5.5, and more out of my computer, but the whole idea is I'm trying not to be sitting at the damn computer all day. Fable 5.2 just determined another session is now the "deputy" so I'm wondering what that means.
I'd pay twice as much per token if there was a guarantee they weren't nerfing the model and if it is as good as the original Fable was when it was released. I'm doubtful though. Whatever "tweaks" they've been doing (safety? performance? "saving the best for ourselves") has really taken away the magic for me. I want the LLM magic back.
I tried Fable 5.2 tonight asking if for a code review of an in-browser machine vision project done by Opus 5.5 over the last 11 days. It used up my 5-hour token budget in half an hour and reported a lot of problems that actually seem pretty bad. But that's just a first impression.
When Fable 5.3 comes out and you give it the same prompt it will take 10 hours , cost twice as much and still find pretty bad problems. I say stick to Opus...
Scary stuff! You should immediately get another subscription and ask another instance of Fable 5.2 to review the findings (don't read them yourself, it will introduce bias in the super AGI's output)
Your first impression is that you have no clue if any of its findings actually mean anything because you can't even understand them or the vibe coded slop you've been creating? Why would anyone care?
You don't seem to have understood my comment very well.
I understand what the findings say, except for one comment about parabolic vertices. What I don't know is whether any of them are true. It could be hallucinating!
But it probably isn't. The bugs are probably real. The computer vision algorithm result images it's showing me are probably real, not fake. Its claims about the different approach to take might be mistaken, as opposed to fake, but even that seems unlikely.
We've gotten to the point where you have to be doing some seriously ambitious stuff to see a difference between fable and opus. Both execute beautifully, but I think for my workflows I'll still be using opus 5.5
Hi everyone, we are developing this new computing platform www.energysimlab.com , would be happy to get your feedback, ideas and collaboration opportunities. The access is currently free for the users.
We are developing this new simulation and computing platform for R&D and tech consultants. Please check it out at www.energysimlab.com (a no-code solution,Feel free to give your opinions, give feedback, and collaborate on more tools)
Fable 5.2 may finally be so far ahead of my intelligence that it's not useful for day-to-day tasks. I gave it something to do in a personal project of mine, and took my standard flow of "let's talk about this, explain the options, tradeoffs, blah blah blah" and after 5 minutes my vision was blurry and I was doubting if I even knew how my own project worked. Gave it to Opus 5.5, and instantly got back on track.
concerning that the models are becoming much better at evading monitors by controlling their visible reasoning, even in deliberate sandabagging (its 25% attempts only).
We need more faithful and generalizable monitoring methods.
Is it only my impression that in recent model releases every vendor is zooming in on hard skills: code, code, some math + physics ? AI is increasingly Jagged ?
We were promised AI can find cure for cancer and in powerpoints most enterprise value comes from Business-AI.
It might as well be that before business-AI can be done we have to cover the basics at superhuman level. All basics, not just entry level jobs like excels, coding jira tickets.
One-shotting a full game (albeit super impressive) looks like overfitting one skill over explaining why we have not much progress in other areas.
I live in an IT bubble and cannot tell if outside IT/math the models are also getting significantly better ?
Interesting to see that with each new release there are more benchmarks where they report results on Mythos instead of Fable. For example, this time scientific research is reported on Mythos rather than Fable, showing one more domain where everyday joe can never use these models.
126 comments
[ 2.8 ms ] story [ 15.9 ms ] threadMy hypothesis - OpenAI and Anthropic are locked in battle. Whoever takes a step, the other one mirrors.
Based on this hypothesis, I think we can expect to see a new Astra model soon given the new Fable release.
And it could be that Anthropic released Fable 5.2 and kept 5.5 in reserve, in case the new Astra model performs better than 5.2
Or maybe 5.5 turned out to be kind of disappointing and they’re going to just go for like a 5.6 or 6 next instead.
Look at OpenAI, they've been outclassed by Opus 5.5 two months ago; obviously they don't yet have a response.
Now Anthropic is releasing this lackluster Fable 5.2, a week or two before their IPO.
It's barely an improvement at all, makes exactly the same mistakes as Opus 5.5. Not even worth using with the speed and cost penalty.
The much rumored internal Anthropic model is listed on the Fable 5.2 system card as an even smaller improvement over Mythos 5.2 than Fable is over Opus 5.5.
I expected a larger version of Opus 5.5 would be a significant improvement to launch before the IPO. Turns out it didn't scale and the improvement is marginal.
I am quite disappointed.
Who knows why they are saving it. Maybe it needs more training and alignment? Maybe they're timing the media attention around the IPO.
https://aibenchy.com/compare/anthropic-claude-fable-5-2-high...
In practice they might recover from those errors themselves if they notice it, so they would have better results, but the tests are mostly one-shot and they can either get it wrong or right, no 2nd tries (apart from the agentic ones and a few coding ones which have a limited number of turns available too).
That being said, the pricing and speed are accurate, regardless of the test results.
Your first impression is that you have no clue if any of its findings actually mean anything because you can't even understand them or the vibe coded slop you've been creating? Why would anyone care?
I understand what the findings say, except for one comment about parabolic vertices. What I don't know is whether any of them are true. It could be hallucinating!
But it probably isn't. The bugs are probably real. The computer vision algorithm result images it's showing me are probably real, not fake. Its claims about the different approach to take might be mistaken, as opposed to fake, but even that seems unlikely.
But I don't know.
We were promised AI can find cure for cancer and in powerpoints most enterprise value comes from Business-AI.
It might as well be that before business-AI can be done we have to cover the basics at superhuman level. All basics, not just entry level jobs like excels, coding jira tickets.
One-shotting a full game (albeit super impressive) looks like overfitting one skill over explaining why we have not much progress in other areas.
I live in an IT bubble and cannot tell if outside IT/math the models are also getting significantly better ?
The improvement is marginal, it makes the same exact errors as Opus, but slower and much more expensive.
From what I saw, it has slightly better taste in some 3D mechanics, but mostly doesn't even bring discernible visual improvement.
GLM-5.3 almost equal even I've tried it and it is proven for security audit in the blockchain realm