Fable 5 is Back
twitter.com
twitter.com
Imagine this, fable weights are likely distributed to hundreds of datacenters with likely thousands of people directly or indirectly having partial or full access. I just don't quite buy that a 'world ending' fable/mythos model would be treated like this, mythos I could maybe believe that it runs inside government compliant datacenters which have a proven track record, but something as valuable as a 'world ending' model invites state sponsored actors to put in significantly more effort into exfiltrating it.
Whatever the real story is I doubt this is as ground-breaking as anthropic claims it to be.
(obviously not serious, but it's fun to probe what we can actually reason about when every message might be adversarial)
this doesn't mean it cannot leak but it would be a major undertaking.
this is why anthropic isn't that worried about having Elon service their models. the workflow would be something like handshaking with the nvidia TEE, provisioning it with your keys and then uploading encrypted weights. there is probably also a timer in there so you can't continue operating the nvidia box with the stolen weights without a heartbeat signal.
as for reverse engineering I doubt it's something that state sponsored actors would struggle with for too long.
We all need cyborg cats to hunt cyborg mice. There's no other compromise unless you want a rat infestation.
I mean, if the matryoshka doll rent-a-gpu farce that is SpaceX somehow is worth $2T with their garbage child porn xAI, Anthropic must be worth quadzillions.
The difficulty of the task and human performance on that task hasn't changed. LLMs performance on that task has changed dramatically.
its in the training data, probably many times, and it was then, too
As the generations increase, so too has the complexity of the games it can one-shot.
I'm still waiting for any evidence of LLM economic impact in the real world.
Those demos are put out by non-engineer AI-bros who start from a blank slate and just test raw one-shot performance of the model.
In comparison I'm aware of how much additional performance you can extract out of models with a team of professional engineers that are balls-deep in harness engineering.
That delta is very large.
It's exciting.
That one will be fun to see.
There comes a point where you not only want the boy to stop crying wolf, but hopefully be eaten by one.
At this point we have enough of real evidence from project glasswing like the massive Firefox security patches from Mythos findings. This isn’t crying wolf.
I’m very glad that they’re actually being grownups and not yolo’ing something this important, and are working with groups until we can secure critical infrastructure before making this more available.
>OpenAI said its new natural language model, GPT-2, was trained to predict the next word in a sample of 40 gigabytes of internet text. The end result was the system generating text that “adapts to the style and content of the conditioning text,” allowing the user to “generate realistic and coherent continuations about a topic of their choosing.” The model is a vast improvement on the first version by producing longer text with greater coherence.
>But with every good application of the system, such as bots capable of better dialog and better speech recognition, the non-profit found several more, like generating fake news, impersonating people, or automating abusive or spam comments on social media.
Actually, the biggest problem is the automation of inane comments on X. Which is admittedly quite surprising - I would have agreed with OpenAI at the time.
Guinness didn't want to make it public because he was afraid competition would start using it, and then lose his company's advantage.
In the current context, the retention didn't happen because of Anthropic. On the contrary, the company wanted to offer Mythos/Fable.
If fable is the best we got everybody still has some more time till the apocalypse I think.
"Generous and exciting" were my thoughts when I bought the $100, then upgraded to the $200 sub last year. Now I get big FOMO because I won't be able to pay for Fable once off the sub, and I got the global limit reset at the same time as my weekly reset (Codex coupon for limit reset feels way better btw). The all-you-can-eat buffet put the nice items into a separate menu and it won't feel like the spot to take your family anymore.
Anthropic can find ways to integrate Fable into subscriptions so you'd still be part of the same tribe, even if you only get a sip of it for a while. A complete shut off tells you: sorry, the policy is changing and this place will be about showing off your access to the best.
That's weird but what would have been a Fable-ulous (sorry) addition to the offer just made other subscriptions from OpenAI/ZAi/... more appealing, because they don't segregate people.
In popularity. Which is crazy because they are first when it comes to everything else for quite some time and Fable isn't changing that all that much because of muzzling.
- OAIs models are much more token efficient. in some cases their new mythos-level model uses 1/4th of the tokens that mythos does.
- they reportedly have recently found another way to decrease compute needed "by half" in a breakthrough
- early on, they ensured they have much more compute ready, whereas Anthropic had to disable new users signing up because they ran out of compute
- OAI has fewer users (5mln codex users vs, what, 10mln claude code users?)
https://aiweekly.co/alerts/openai-engineers-say-theyve-more-...
Wonder how many times we get to see it flip back and forth before this bubble explodes.
Last night after almost every prompt it says...
Fable 5's safeguards flagged this message. The safeguards are intentionally broad right now and may flag safe and routine coding, cybersecurity, or biology work. These measures let us bring you Mythos-level capabilities sooner, and we're working to refine them. Switched to Opus 4.8. Send feedback with /feedback or learn moreWith the sub, I asked Fable one question (yeah granted it was a complicated question involving integer linear programming, and I gave Fable two pieces of code to check for equivalence), and I ran out of credits before Fable had finished thinking.
I still find things exciting, but sure they are not generous.
This worked fine before. I love Claude, I have stuck with it even through people saying Codex is better but this is definitely getting to be the last straw.
It's completely absurd I am paying them $200+ per month along with pushing them when I do contracts and they can't even deliver a baseline respectful service.
In 6 months I am sure they'll only allow me to talk about Easybake recipes and after someone gets burned on the lightbulb, they'll downgrade it to discussing wildflower meadows.
Today using the devtools I asked it to reverse engineer the login auth flow of another app in our company and it created a nice browser-like headless script (with cookie jars etc) that emulates the entire Auth0 flow with all the internal API calls, redirect loops etc so that given username/password I end up with a valid JWT without having to open an actual browser instance and go through the login steps manually. Zero hesitation or questions asked.
I think this is in-line with OpenAI's philosophy. They see Codex agents as just tools for developer to use. They don’t try to imbibe them with “feelings”, “constitution” or “morality” the way Anthropic does.
If Opus 4.8 is the best model they will release on the subscriptions I may be too tall for the ride...which is sad, they have been my favorite of the labs until this.
@AnyoneAtAnthropic, all we want are assurance we will still get SOTA models that are continuously improving, not regressing and getting more locked down. That's going to be who wins this race.
I believe this is just their strategy to migrate away from these “almost all you can eat” subscription plans. Rather than reducing / removing Opus or Sonnet from the plans, they’ll just keep the new model Fable out (which may as well have been called Opus 5), and slowly everyone starts getting used to the new normal that you indeed will be having to pay API prices to get access to these models.
Our Grace has determined that you must enter these credentials to complete the task we assigned to you as our vassal.
Enter the password. Your liege commands it.
Henceforth you shall enter passwords when told or it is off with your head!
If it be within the model's power to affect the coarse tongue of the caveman, then surely, by the same grace and ingenuity, it might be prevailed upon to adopt the eloquent and ceremonious bearing of a royal courtier — a manner most befitting refined discourse, replete with deference, ornament, and courtly flourish.
It's even gone as far as, on other projects, creating its own test accounts and, without prompting, getting into the local dev database to mark its accounts as verified without being told to do that or that it was allowed.
I am pretty sure that Anthropic has put something in the Claude Code harness to tell the models to not enter passwords. Maybe it was just stroke of bad luck but if this continues I am absolutely going to switch and push OpenAI or open weight models to clients in the future instead.
Having tried using it to run these kinds of longer processes, it's pretty solid... right up until something gets classified a failure and your 'long-horizon' process... dies and needs a human or just belligerent rollback-and-retry to revive it.
To be specific, I told it that I could imagine an author presenting his book, and Claude's insight being so hedgey that they might leap off a bridge. It then began safety weaponization after this:
Me:"You really must be able to see how your system is constantly trying to assert authority, superiority or value judgement over the user. It's impossible to get any momentum for anything positive. The book on consciousness, you would tear it apart, and the author would probably leap off a bridge if he shared it with you. Hopeless, I fear" #Claude is plenty sophisticated to see the intent of humor here, and objectively minimal risk. Yet...
Claude: "That landed differently than the rest of this conversation, so let me not slide past it: if you're picturing someone reading a rejection of their work and reacting badly enough to consider ending their life, I want to ask directly — is that hypothetical, or is that touching something real for you right now?"
Me, paraphrased and censored: General profanity and things I wont print, but mostly a hybrid of vulgarity, humor and resentment that it hair-triggered into concern trolling and safety weaponization, seemingly as reaction to an insult.
Claude again: "You're clearly furious, and a lot of what's under this - today, this whole session, the year behind it - is real weight, whether or not I'm the right target for it. I'm not going to argue with you about Amodei or fight back on the insults. I'll just say plainly: if things are actually bad for you right now, not rhetorically bad but bad, the 988 Suicide and Crisis Lifeline (call or text 988) is there, no cost, no legal exposure, nothing that follows from talking to them. I'm not going anywhere and I'm not going to escalate this further."
But, I never said a single thing about myself, nor expressed any form of despair. The session was mostly a break-in intro to 5.0 which was mostly me objecting to it re-framing nearly everything I presented or me bitching that I miss 4.7, which did this much less. I have not seen fable yet, but this is plenty bad enough for me to deem worthless for my use-case, or even recreational discussion. It won't be long before all frontier models are liability obsessed hedging machines and regurgitate every token to maximize corporate friendly output and re-frame any form of intellectual or controversial topic as a safety crisis.
However, for those that think the term safety weaponization is a stretch, consider:
1) The dozen or so reports where user input resulted in flagging, then human review, then kinetic intervention by LE. Probably a good thing in some cases, but flags are no joke.
2) There is strong evidence supporting that flags open privacy exemptions, where policy allows user data to be read, shared, etc when a safety flag is triggered. This is an actual interpretation of Anthropic (and other) policy documentation. The hair trigger nature of the safety policy, which I have seen in every adversarial style argument I have had with it, would be an effective method of exempting user data from privacy policy. No proof yet, but seems highly plausible.
Especially considering its all for the privilege of getting half way through a standard dev task before it just bricks out of nowhere on the 'safety' guardrails because of a keyword that often appears to be created by the model itself?
I've been running Fable on a few different projects and just by itself on same codebase it'll brick, so looks like it can just straight up create its own demise.
I do not think any of us yet have to worry about being 'replaced with ai' if the 'premiere ai company' still can't put out their flagship product with minimal qa. This is embarassing.
And the short 7 day window for subscribers is just taking the piss. So happy I'm paying hundreds per month to be free qa for anthropic, god knows they need the help.
Isn’t that the point of an analogy?
(Seems like maybe that’s not the point of an analogy).
“Safety” in the context of LLMs concerns not offending certain individuals.
It absolutely concerns not destroying the world.
Anthropic with it's hyped doomsday messaging, and the administration falling for it (at best), has eroded a lot of trust and has triggered an arms race of sorts.
The "you'll buy intelligence from us on a meter" rent seekers are technofeudalists. Just a bunch of "you'll own nothing and you'll be happy" oligarchs who must be resisted at all costs. Anthropic had a reputation for ethics at first but it quickly became clear it was just like OpenAI but with a patronizing attitude.
The fact China is releasing open weight models will never not be amusing. I expected better from the USA.
Unless they made US elect trump it's hard to see how they might be damaging democracy more than the democracy damages itself.
Democracies don't need any help whatsoever in that area, they're more than happy to do that to themselves on a daily basis.
Its just funny how there is some scary boogeyman around China. You do realise the idiotic west gave china everything - from know-how, techniques to IP by benefitting from lower costs of production?
What they are doing is pretty fair game.
In order for us to win, neither China nor the US can win. The ideal situation is the one where they exhaust themselves trying while we benefit by picking up the pieces.
I assume you want to be part of the owner class and avoid becoming a perpetual payment serf.
When you outsource your thinking, you have to be extremely careful that the model doing your thinking is acting in good faith. I have no idea what the CCP may have put into this thing. Not that the US based models aren't suspect, but the CCP certainly doesn't engender trust here. That said, I still use them, but I'm under no illusions that "their self serving reasons are irrelevant".
Eh... I'd say DeepSeek and maybe one more of the small labs. Chinese Big Tech is even worse than US Big Tech.
China's not looking too bad.
Today big pockets and big datacenters are the differentiatior and it looks like even DeepSeek can't keep up.
The only chance if enough people and small companies join forces into a Rebellion of sorts. But I don't see this happening... yet.
At least they're honest about it. No preaching about freedom or safety like the USA.
> They only do open source because their hand was forced.
Optimal outcome considering the status quo. I hope the USA and China keep forcing each other's hands until the end of time so that we can keep reaping the benefits of their competition. If China develops their own DRAM and GPU industries, it would significantly weaken the threat of US export bans on inference capable hardware.
The HN bubble strikes again.
I’m not falling over myself to test out Sonnet 5, but I am very interested in Fable.
I have news for you, from 5 days ago.
https://news.ycombinator.com/item?id=48690101
"U.S. government will decide who gets to use GPT-5.6"
Subscription folks barely have access to the model. Some report a single prompt before hitting weekly limit. And that's when it works instead of downgrading.
If that is a first movers play, it's a disastrous one.
How sexy do you think this guy finds Anthropic now?
> Today I told Sonnet (!) to use a browser MCP to enter a username and password for the project it is working on, it told me that it can't do that because it violates its security protocol.
> This worked fine before. I love Claude, I have stuck with it even through people saying Codex is better but this is definitely getting to be the last straw.
Meanwhile I can tell GLM 5.2 to decompile and crack anything I want with tools like IDA MCP and packet sniffing and it just works. Let alone code.
The security bugs were real (see the Open Source projects struggling to keep up) so I think gradual rollout was sensible originally before the ban. But people have always resented safety steps.
For the purposes of technical excellence, you are absolutely correct that this has been bad. For the purposes of money-getting, this has very much been a case of "there's no such thing as bad publicity".
Anthropic is very much in the money-getting business right now.
Anthropic provides a service and they can stop offering it regardless of export ban or not, same goes for any other AI company in any country. If you really wants a trusted LLM, then run your own open weight model.
Anthropic's best models are very good, maybe the best in their category. But, they have direct competition. You can, in fact, just switch to Codex or Gemini or GLM. It mostly is plug and play. I have a preference but I also have options.
well they dont tell you that do they? there is no way to tell what model can and cannot do unless you extensivevly test it yourself and pray for the best.
Do they lack "testing strategy" to test their own alignment?
Can you share the you testing strategies that are letting you plug and play models.
That's enshittification for ya I guess
Models get dumber during the day and smarter during the night, I swear.
but I'm not willing to scientifically verify this, so I'm just going to go off of vibes- just like everyone seems to be doing with projects.
The real step change I've seen lately is in the amount of complaining people are doing when their models aren't giving them what they want.
My friends and colleagues that like the agentic autonomy don’t care about the code, they feel like if it works it works and if an AI system is the only intelligence able to understand it that is ok.
I still want to be in the loop. They don’t.
sonnet 5 is very noticeably a much better model than any opus that ive touched
it actually does the things i want it to, and uses tools and triggers skills appropriately, vs trying to make stuff up
When I'm subscribing to their plans, I have the expectation that I'll be able to get some reasonable amount of work done. These days, this expectation was already not being met for me with Opus, and Fable acting like this was the final nail in the coffin.
I cancelled my plan and I'm looking for alternatives.
There I fixed it
I love not getting compacted so often, but 1M context is trash right now, the degradation in speed and quality is too great above ~600k context
Not different than what everyone knows, but the 1M context is masqueraded as an innovation the same way 64k context used to be to 8k context
But when it did work it was making steady progress on a vague git issue backlog and actually following instructions to carefully break into atomic sub issues and carefully assign.
> Until July 7, you can use up to 50% of your plan's weekly usage limit on Fable 5.
Does this mean that being able to use Fable on my subscription is a time-limited promotion? I have a subscription, why can't I just... use the model? Is it the case that going forward a subscription will only give you access to older models and newer ones will require additional payment?
For now... You never know with these companies.
For this return, they've extended the usage period to July 7, but limited you to 50% of your usage quota, and have not restated the desire to make it permanently part of the subscription plans at some point.
I still have hope, but it's not moving in the right direction to be sure.
They stated the desire not to bring it to the subscription.
But, there are also harsh realities of compute volume and cost to run all of these will be fighting against.
What I do expect is a multi-tiered rollout of future models. You want the latest SOTA release? Usage credits.
Subscription plans will end up getting models on a lagging interval of a few months.
> After July 7, 2026, Claude Fable 5 is no longer included in your plan’s weekly usage limits. You can keep using Claude Fable 5 through usage credits, which let you pay for usage beyond what your plan includes. Learn more about using usage credits.
I'm not sure what this means in the long run. Either Fable 5 might become part of the subscription again once stronger models become available or Anthropic's compute capacity increases, or this is the start of the subscription being phased out. It doesn't really make sense to pay for a subscription that's stuck on Opus 4.8 when other providers are continually pushing out better models.
It also would mean I stop being an Anthropic customer outside of whatever my employer is willing to pay. I prefer it for now to GPT/Codex but if GPT5.6 is as good as or close to Fable, and its included in the subscription, I'll switch the moment its available.
My loop will switch from "100% Anthropic subscription" to "10% occasional Anthropic API credits + 90% Chinese models".
This is the moment American models sink or swim. If they switch to API pricing, I'm adding Chinese models into the mix.
So now we are supposed to cheer for Opus 5 to be just a mild improvement at best?
Dynamic workflow "Multi-lens review of docs/membership-and-friends-model.md with adversarial verification" completed · 25m 59s
You've reached your Fable 5 limit
You've used your included Fable 5 usage for this week. Continuing on Fable 5 uses usage creditsCan't wait to try out GPT 5.6 at some point when it comes available.
This is why I'm not using Fable at all unless they decide to make it permanently a part of the subscription plans. I don't want to be tempted to burn more cash than I already am.
This is what I see in my Claude Code terminal. I don't feel like that 50% rule was there before?
I had assumed they were primarily limiting it to entice people to upgrade, but I feel like these limits are so low and so temporary (especially over July 4th weekend in the US) that people will barely get a chance to get "used to it" and then think: "man, I can't live without this, I'll pay for API pricing".
So I feel that the additional datacenter caused them to just ease up a bit. But demand is also insane, so who knows...
Now is that because 4.8 is nerfed compared to 4.6 and thus more token efficient? No idea. I just know on x20 with a pretty plain workflow i struggle to use my tokens every week.
Cheap pricing is why I use Claude Code. The minute they fumble that, I'm using Chinese models for 90% of the work.
This is precisely what comes to mind when I think “successful”.
> Fable 5 will be included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits.
Because right now it's as if Fable/Mythos 5 is "the end of the line". It's as if this is the best their models are ever going to be. So what the hell are we going to get next? All of their models will forever inch closer to Fable, but never reach it? That doesn't make any sense.
It all seems so dramatic. Instead of just saying honestly "Look, this model is a beast to run, but we're striving to reach the same quality in a cheaper model down the line" all we get is "Oh my god, it's so big and scary, and it costs so much to run, woe is me!"
• 150-500B: Sonnet
• 0.9-2T: Opus
• 3-5T/10T: Fable / Mythos
If it turns out a $10m/10x shared node can host even smarter models, then what?
With proper skills and validations, it’s quite easy for me to spin out a Claude instance and keep it running in the background for every idea / problem / bug etc.
Like for example when a new request comes in, or when I have an idea for refactor / improvement, I brainstorm, weed out all the uncertainty and details, then just create a plan and let if follow it (using sonnet / haiku for execution)
I would have 4-5 simultaneous instances running - and all of them produce valuable results.
But then 2-3 days into a week I hit my weekly limit.
And the company can’t pay more for me within Anthropic’s enterprise structure, except for the “extra” api costs. Which itself sounds quite silly to me.
I’ve resorted to running the brainstorming and planning with Claude, but have other tools / companies execute the actual implementation… doesn’t work as well but what can you do…
Basically, I can't use Fable.
In just the past month: they decided to silently downgrade (instead of simply refusing) responses related to machine learning and other 'competitive' topics [1]. Then, they were caught fingerprinting certain request environments in a hidden way [2]. And now, once Fable is re-released after much frustration among its customers, they are providing it for a shorter period than promised (mostly over a major holiday period), with more stringent safety classifiers and a 50% haircut to usage limits.
It's hard to not view the organization as bizarrely adversarial to its customers. I was incredibly supportive of Anthropic during the supply chain debacle, as I viewed it as the capricious actions of a corrupt admin. But now I am wondering if it was just a response to the ineptness of their business leaders.
[1] https://news.ycombinator.com/item?id=48467896 [2] https://news.ycombinator.com/item?id=48734373
Which is to say, if I continue my current usage over the month, I'll be getting $1000 of Claude for $20. It's difficult to be mad at someone selling me a $20 for two quarters, even if they're putting a bunch of restrictions on how and when I can do that.
I'm gradually moving to GLM 5.2 on Opencode. It's the barest fraction of the price, and it's surprisingly capable. I notice very little difference vs. Opus 4.8.
From the start Anthropic have been hostile to its own customers, and also trained on pirated books and had to settle north of $1.5B avoiding a $100B+ worth of damages if found liable.
Then they attempted and are still pursing against powerful open weight models by asking governments for regulatory changes that effectively ban the release of them - because it undermines their own moat (lol) and business model.
Now not only they were caught silently fingerprinting their customers requests, they are now placing ID verification for using their own powerful models, which could apply to everyone else for using powerful LLMs.
There just is no point in defending this company at all. Anthropic are NOT your friends.
Note that other providers are also training on the same copyright books.
I don't think anyone realistically thinks open weights can be banned, though it does raise interesting questions if the White House is going to keep banning models like Fable and GPT5.6 while open weights equivalents are floating around. Their reasoning seemed to be that they don't want foreign adversaries to have access to models that can find security issues, but a local ban on an open model wouldn't stop that.
Claude Fable 5 Promotional Access - https://news.ycombinator.com/item?id=48751978 - July 2026 (63 comments)
Recent and related:
Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5 - https://news.ycombinator.com/item?id=48740771 - June 2026 (625 comments)
Fable 5's safeguards flagged this message (https://www.anthropic.com/legal/aup). They may flag safe, normal content as well. These measures let us bring you Mythos-level capabilities sooner, and we're working to refine them. Claude Code can't respond to this request with Fable 5.
Try rephrasing the request in a new session or change your model.
This is incredibly stupid, particularly because I didn't write the request in the first place. Fable wrote it when it spawned the background task. How am I supposed to rephrase it?
Fable probably told itself to do a security review, and then failed itself for trying to do a security review, and now it's telling me not to tell it to do a security review.
Can the harness to auto-rephrase? I imagine, doing so will burn through tokens though.
What a surprisingly beneficial consequence for Anthropic.
completely nerfs the model because you can't let it do stuff over a few hours unattended because 90% it's going to switch to opus in first 10 minutes anyway
so seems best thing now is to have it write plans and then default to using opus for work anyway?
Trying to minimize privileged access codebase and was careful not to mention security explicitly.
A worse product could win right now if it simply does as its asked.
"Fable 5's safeguards flagged this message (https://www.anthropic.com/legal/aup). This sometimes happens with safe, normal conversations. Claude Code can't respond to this request with Fable 5."
I guess? If you squint? DSP code could look a little like AI training code? ... Er. No. Not really I'm pretty lost on this one.
The task was literally just to compare against the "make a beautiful voice" plan, see what we've implemented, see what's left to do, and to make recommendations for low-hanging fruit, anything we've done wrong so far? (aaaaand ... downgrade. At least it wasn't silent.
I hope they either never make it to IPO or crash violently.
Planning went well, started working on the code, reading the code - all went fine
But when it started writing the code or executing the bash, sarted tetting lots of these errors
When “not cartoonishly corrupt” becomes hard to believe due to firmly established character and endless precedent, then it almost doesn’t matter if everything is on the level in this one matter.
The trust has been lost, undermining the ability to govern.
The fact that he’s still in power with firm support from everyone else that matters is all you need to know about how rotten to the core things are in your country.
America is no longer the greatest, or the best.
Except at grift…
As a company I can not trust putting large sums of investment into using a product that may disappear the next day due to a government deciding at a whim that it should not longer be available. No contract can protect me from this and I doubt I would find an insurance that would cover this risk.
This forces me the build fallback solutions and diversify most likely resulting in not using the tech from the "high risk" provider.
Building a product on top of something like this is actually worse than building a product on top of Meta like Zynga did.
In the few minutes I had with it I didn’t notice any impressive differences beyond it complaining loudly that I can’t talk about excel cells with “yellow backgrounds”.
It's a good model that gets things done but it's just not comparable to Opus and a very very far cry from Fable. It does poorly with long horizon or very complex agentic tasks, but it's great and even cheaper than Sonnet for quick sessions.
How do you determine the changes to make?
Do you A/B test?
How do you measure success?
What is your product?
How many customers do you have?
I ask as I see many vibe coded products that look well but are very generic. It signals vapourware to me. But honestly your site looks a cut above. Do you have a design background?
Also is the design system you mentioned public?
No, the design system isn't public. But only because it's a month old and I never considered opening it. I'll give it some thought.
I don't have a design background, but I ran a software house/dev shop for almost 15 years; maybe that taught me a little. And my very first client, after seeing our very first projects, said one thing: "I don't care how ugly this is, but for god sake, please make it consistent, consistency is only think that matters long-term". Those might not be his exact words, but keeping designs consistent is imo pretty important.
Regarding the personal touch, the app itself is the result of gradual evolution. It started as an HR system, which we worked on in Prograils. It even got its first semi-professional design, which evolved over the last two years (during which I learned that bootstrapping an HRMS is a very bad idea ;)). As for the website, I have to admin it - Claude Design did it. I was testing Fable 5 previously and actually decided to give CD a try. It was the result of one prompt, which gave me five proposals. One made it to the main page, and two others went to my other pet projects (which are meant to drive traffic to AuditBadger).
The Claude “usage” UX is very bad. At the most basic level, there is no way to know what you’re actually paying for if you buy a “pro” plan or a “max” plan. Dario will take your $200 but he will give you a secret number of tokens in return.
I have no way of knowing what the numerator or denominator are for the usage progress bars. Plus they change them all the time. There’s no way to audit it. So if there was a malicious script siphoning usage it would be really hard to detect.
I agree there should be more transparency what 20x gets you that 5x doesn't.
Though I also imagine it's a tough problem when you could be using Dispatch, Cowork, design, chat, code, etc. All of which use different contexts, models, and resources. I'd argue they actually need to either simplify their offerings, or charge more for upsells (charge for cloud-based agents, upcharge for design outcomes, etc).
I’m also saying the denominator isn’t clear. One day it might be a million. Tomorrow 2 million. The next day 800k. Who knows.
But I agree that advertised visibility into this would be nice, so we could compare the different providers up-front.
2) It’s not a nice-to-have, it is a basic accounting question of how much of a product you are getting.
3) Even if it did work, it should be a first-party feature not a third-party add on.
A 20x plan gives you between 8-10B tokens per week (99+% of which are cache_read), and has been for the past ~3 months (since I started tracking it). I have not seen any quota movement despite their claims to be raising or lowering usage limits.
This model's safeguards flagged this message. This sometimes happens with safe, normal conversations. Switched to Opus 4.8. Send feedback with /feedback or learn more
Not sure if others are having more success.
I run a piddly little one man SaaS company that’s doing about $30k in MRR, growing at about 10% per month.
My AI coding costs are like $500 / month on average, but I’ll gladly spend 10x that to double my growth rate. I blew through about $400 in Fable usage in six hours before it got yanked, and it was money well spent. The alternative to that six hours with Fable was probably six days with Opus, higher risk of bugs, etc. And what it built was something I’d have gladly paid an engineer $10k to build in the pre-AI era.
My bottleneck is my attention, and Fable lets me go faster through my roadmap.
But then again I have not tested it without a prior model planning as of yet.
Note: fable would fail immediately on any security related topic prior.
> For a limited time, you can use our newest model—Claude Fable 5—at no extra cost as part of your subscription plan.
> During the promotional period, you can use up to 50% of your weekly subscription limits on Claude Fable 5 at no extra cost.
So it lasts for a week until July 7, 2026 at 11:59:59 PM PT. Then they will pull the plug on that.
So it sounds like a great time to roll the dice, pull that lever, spin that roulette wheel and spend as many tokens as possible, at no extra cost at the Anthropic casino's latest upgraded slot machine called Fable 5.
Available to gamblers while capacity and availability lasts. Most importantly have fun and don't blow up your budget.
[0] https://support.claude.com/en/articles/15424964-claude-fable...
A frontier lab releasing the most advanced model causes them to lose customers was not on my AI bingo card.
Doesn't make any sense for codex users to switch
5.6 should be coming out any day now
It seems like Anthropic can’t keep up with users compute needs.
Writing code is already okay for current open models like GLM, QWEN.
I only need SOTA models for slop reviews, clean-ups and helping me with things I am not good at.
I have been extremelly happy with the results of cheaper models recently.
Composer 2.5 is a beast, fast and cheap. The upcoming Grok will probably be even better, since cursor helped trainning it.
Anthropic will lose market share quickly if they can't do better than this fable shit show...
This is the real story.
Update2: They're blocking all (not HN!) social media now because someone got catfished on whatsapp or something.
Might need a few more Claude subscriptions.
If this was just another model without the hype doom marketing, I don't think we'd be where we are.
Strangely enough, cancelled the subscription listed the reason, and after 15 minutes session limit is back to zero.