$100 just to make 3D models? Of course not. But $100 per month for tools that can help create plugins, dashboards, reports, websites, and some usable 3D models? That becomes a very different value calculation.
Yeah. Exactly.
Guys, you still don’t realize that the prices you’re seeing now are in no way the final prices this thing will actually have.
You cannot create “multiple” whatever with 100$ of API tokens when all of this will stop being massively subsidized at a crazy loss ratio and you will have to pay the actual cost.
At Fable’s tokens billing/consuption, with 100$ you will be able fix a couple of bugs on already existing code (if you barely know what you are doing) and that’s it.
I’ve explained this several times, but you seem to be having a hard time understanding it.
So I’ll give up trying to explain it further.
Good luck. 
The AI naysayers have been very loud recently and to some degree I can appreciate their arguments however we keep seeing these things do stuff that we never thought possible. In other words the dominoes keep falling.
I still remember when Deep Blue beat Garry Kasparov at chess, not one thought that would ever be possible. If you were to ask me in 2021 if I ever thought an AI would help me write code for my plugins I would have hung up the phone or laughed in your face. AI doesn’t write my code autonomously just yet but if I corral it just right it can definitely take care of some of the more tedious tasks.
Yes, the financial situation with these AI companies is probably not sustainable and at some point we will probably have an AI “crash”, anyone remember the internet crash of 2000 or what we called the Dot-com Bubble. Sure there were some corrections in the markets and trillions were erased and hundreds of companies went belly up, but at the end of the day the internet itself stayed and eventually flourished.
If we endure a AI crash I suspect the same thing will happen again, companies will be restructured, IP will change hands but in the end AI is here to stay, and it will continue to improve. As AI improves it will only get better at what it does and be more widely applied throughout society. It will find its way into every nook and cranny where it makes financial sense. If it increases efficiency and lowers costs then why would you not use it?
All these guys that do graphic design and movie making (ie. CGI), the writing is already on the wall. Programmers, designers, architects and engineers, our days are also numbered.
Once we invented the tractor and they matured slightly (ie. the John Deere 2010 era in the 1960’s) did we ever go back to plowing our fields with teams of horses or ox? Would we ever go back to our pre-PC days when we drew up our plans on those beautiful inclined drafting tables?
Who knows what form AI will take in the coming years and how it will be adopted in every industry but one thing is for sure it will change our society for better or for worse.
If you really think about it we are in an AI arms race right now, both nationally and internationally. Everyone is afraid that the “other guy” is going to create the ultimate AI god or weapon or both. That alone will continue to drive this thing and unfortunately it will cause us to discard any safety limits or precautions that we should be paying way more attention to right now.
The release of Mythos was a good example of this. Anthropic knows that this engine is too powerful and too capable to the point that in the wrong hands it can do some serious damage. However market pressure forced them to release Fable 5 to the public, a slightly weakened and guardrail enforced version of Mythos. However, if OpenAI declares code red again and pushes out an even more powerful ChatGPT then you can bet that Anthropic won’t take it sitting down. Eventually extremely powerful AI engines will be in the hands of good and bad actors, you can’t stop progress.
This cycle is already set in motion and will only end when AGI or ASI is finally reached. We are all on an AI train headed toward a cliff and the brakes have been disabled. What that means for humanity is anyone’s guess.
Having an EU passport makes things easier in your case and you should take advantage of it. I just wanted to point out the important distinction about nationality. Because of the EU the Irish can’t catch (most of) their own fish, extract their own oil or gas, or dig their own turf- Now they want to force us to join their army and die for them- which they promised on their grandmothers graves wasn’t ever going to happen if we just voted for more integration 20 years ago. We played our own part in creating that situation, but there you have it. That fine institution of unelected bureaucrats make Stalin and The Small Moustache Guy look like saints. I won’t rant too much. Promise.
This all feeds into the whole discussion about AI. The EU are fully on board with tracking tracing and databasing everything using AI-Digital ID. AI is useful for certain things but a nightmare for others- Research and decision-making capabilities are appallingly bad and getting worse. AI will only work as far as it’s programmers decide, and can turn it off at any time just when we start to depend on it. Then the fun really starts.
I’m not against AI per se. It’s just a tool after all, but it comes with some extremely important caveats which require our close attention.
BTW I wouldn’t dream of telling anyone what words they should use for anything- That would be highly hypocritical and self-defeating. All I’ll do is comment and maybe irritate sometimes. 
I’m not sure I can see much of a difference since Brexit- Though I’m not living there so I won’t pretend to know. What I do know is that the press likes to reinforce certain viewpoints- That said, I can agree with you that Farage is as snake-like as the rest of them. Damned if you do and damned if you don’t if you ask me. At least you don’t have to worry as much about what’s going on in Mordor (Brussels) as I do. But then you have your own problems over there of course. [cough] digital-id [cough]
We hear you - we just don’t agree.
Yes, the prices will go up. In your opinion, they’ll go up to the API prices, which basically make the tool unusable. In our opinion, they will be something more than what is current, but not so high that the tools become worthless to normal people.
You have an opinion on what will happen in the future and ours is different - that is fine - we will all see what happens after the IPOs.
100% this. And honestly, if it went up to $200/month? It’s STILL not that bad of a deal - you have to compare it to the cost of hiring and continuing to pay a software developer.
I think this is a really good way of looking at this. This tech is here, and the world isn’t going back. In some way or another, AI is going to get incorporated almost everywhere.
That’s what I’m saying is both exciting and also scary - it has the possibility of improving a LOT of things, but it’s also getting dropped into a LOT of things.
Yep - they had to release it to try to inflate the demand around their IPO (which I think is overvalued, but I’ve been wrong before), and I don’t love the “guardrails” they set.
“We couldn’t get it to actually refuse to talk about things it shouldn’t, so we set our system up to redirect requests on those topics to Opus” feels like the kind of thing people could get around pretty easily if they tried.
I read a book about the risks of trying to develop AGI called “If Anyone Builds It, Everyone Dies” (a bright, sunny title, I know). I think it might be a BIT overblown (I’m sure someone will be along to talk about how they hate the author), but it also raised very good points about the nature of intelligence, and how we really don’t fully understand what these models are doing.
Even if it’s not “intelligence,” the models do things, and those things could have bad implications for people if they do the wrong things.
I really wish we could slow it all down just a bit to make sure it’s happening the right way, but I agree - I don’t see it happening
If you use these models enough, like I do it is pretty clear that these AI brains have intelligence. It may be completely alien to our own thought patterns or intelligence but it certainly can reason and think very well for itself.
I just used ChatGPT 5.5 Plus (High) to help iron out a couple minor but very detailed alignment bugs with my handrail fitting algorithm for curved stairs (elliptical and circular). Here is my very wordy and not so clear attempt at a description of the problem that I prompted ChatGPT with:
At a very close inspection of the top or up easing on both handrails I did notice that the up easing is not perfectly aligned with the X axis (this is to be expected since we are probably using the vector between the last two points of the handrail). Effectively what this does though is it rotates the final face of the up easing ever so slightly in the XY plane and this is noticeable because the top fitting is aligned perfectly (orthogonal) with the X, Y and Z axis. If we were to rotate this up easing so that it is orthogonal then would this create a gap at the mating of the hand rail and the easing or could we rotate it at the top center mating point, keep the mating face (tangent face the same) the same and only the rest of the mesh is rotated true to the X axis (a very minor rotation). Hopefully what I am describing makes sense. Dimensionally we don’t want to mess up the cross sectional profile of the mesh as it mates with the fitting but I don’t believe making this adjustment will have an impact other than at the one transition at the mating segment between the handrail and the easing. if you need me to explain further let me know, this is probably as clear as mud to you.
Here is the exact screenshot I sent it (so that it had some visual context):
I won’t bore you with the response since it is a bunch of math and vectors and then some ruby code and then further explanations of the intended fix (which worked perfectly as expected).
My point is ChatGPT completely understood a rather complex spatial problem and then reasoned about the correct fix and also discussed the implications this would have on the final geometry (ie. the tradeoffs). It then ascertained that my suggested idea was indeed the correct direction to take and only the segment where the easing mated to the rail would be slightly twisted or distorted due to it being a mitered cut.
You can’t tell me that these AI brains can’t think. Honestly they can already think better than I can and I’ve been doing this stuff for almost 11 years now.
Sounds like you need your own Irexit Le_Corb – but will Ireland make it on it’s own?
Maybe Ireland needs to join the UK 
You have absolutely every right to disagree with me and to have any opinion you want, and that’s fine, of course.
But I’d suggested we stop confusing publicly available billing data and Anthropic’s official announcements with “my opinion.”
Enjoy Fable free in your subscription until June 22.. then it will officially be API only at 50$ per million tokens.
Then they “eventually would like to introduce that again in Max plans”, but yeah.. As you said, we need to see if they commit to that after they go IPO.. that’s where the speculation and opinions start.. so you can totally be more optimistic than I am.
That’s fine, neither of us have a magic crystal ball.
We can only make assumptions based on trends.
And the current trend is: every major company is reducing usage across the board.
@panixia I built my plugin with Claude 4.6 on the $20 plan, before Fable 5 came out. While Fable 5 is impressive for modeling, especially with details like door handles and fixtures, I still think 4.6 is a solid tool for creating SketchUp plugins. Thanks for reminding everyone about the timeline.
Let’s go ahead and put the whole snip in here just for context. Granted, you said it in your text, but it does change the overall contextual “feel” of what they put on their site. Now it’s a question of if they do what they said they’re going to do or not.
I personally would suspect that the amount of compute they have available also affects this rather heavily.
Beyond that - I suppose time will tell
The reasoning strings that I’ve seen in Claude, especially for the bigger models, like Opus and Fable, are fascinating.
It starts by stating the problem, then trying a solution. Sometimes it’ll be in the middle of a solution and stop and say “Wait, no, this is wrong, I need to try something else.” Here’s one of the notes from the Fable 5 floor plan I used as the example in my video.
It definitely does a form of reasoning, which is why I think at a certain point the difference between AGI and a super advanced LLM with reasoning skills COULD be a semantic one. I understand that the actual difference is important, but for 99% of practical uses, it might not be.
I’m excited to see what this can do in your hands, as you’ve got the coding knowledge and experience plus you’re seeing how much this can change things.
My view with the pricing is that its “your first hit is free” once you are hooked the prices get jacked up.
At the moment prices are low to get everyone onboard with the idea and get it ingrained into everyone’s work flow. Just like Uber making massive losses to push out all the taxi drivers, chain cinemas doing rock bottom prices until the competition leave and Microsoft pushing everyone on to 365 subscription. once the competition is gone or you forget how to do it the old way the prices go up up and up.
Yes, I stated it my text.
I wanted to put some really needed emphasis on the fact that the only thing that you can take for granted is that they’ll remove it within 10 dats and that’s official.
The rest is just whisful-thinking/marketing promises of big tech company.
They say “when sufficient capacity allows us to do so”.
Which is a bit like saying, “Believe me, when I get younger, I’ll do this.”
All they did in the last few month has been just cutting down usage in multiple ways over and over again.
They always claimed that the problem was “limited compute capacity”.
They are also removed Opus and are planning to remove CC from the 20$ plan for the exact same “limited compute capacity” reason (and they already A/B tested it).
I don’t expect that with the (way) more consuming Mythos/Fable models and the increasing demand from user they will have some capacity to spare anytime soon.
But of course, you are free to believe that they’ll do.
For now, they’ll remove it and it will be “pay as you go” only.
Once they added the chain of though reasoning and also a memory these things went from an LLM (basically just advanced auto-complete) to an LRM (Large Reasoning Models). LLM is a bit of a misnomer at this point, I really thing we should stop using that term for the latest engines.
ChatGPT (when I first started using it in Aug. 2025) initially could not remember anything from previous chats. Then that changed and it made a dramatic difference with its capabilities and its usefulness.
I am already leveraging the efficiencies of AI to some degree. It can take care of tedious tasks quite well when properly corralled and directed. You don’t want to give it too much latitude otherwise you tend to get slightly unpredictable results, in other words you need to carefully define what you want it to do especially with complex problems.
No, they didn’t. It’s still there (and Fable too - but that’s quite pointless with this plan)
But they seem to be constantly tinkering with the general usage limits, which, according to my perception, can vary quite a bit throughout the day and week…
My bad, I made confusion with Github Copilot Pro plan.
Indeed.
Best thing I’ve ever seen on this forum!
That didn’t take long.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.