Hacker News Reader

Comment by bluepeter | Hacker News Reader

Fable 5 apparently can't be used for coding? (This is from Anthropic's announcement.)

> After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8.

Edit: the above was from their tweet announcement at https://x.com/AnthropicAI/status/2072163884430229756 ... the associated blog post at https://www.anthropic.com/news/redeploying-fable-5 suggests it was just poorly written and coding can still be done with Fable, just with overeager bouncing of "some routine coding and debugging tasks" to Opus.

View on HN

> 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.

> The new classifier also comes at the cost of flagging benign requests more often during routine coding and debugging tasks.

Here's Fable 5, the strongest model. Actually try to use it to harden your code and it turns into Opus 4.8. You have seven days to use it, and only half of that time's worth in actual usage. Enjoy.

Looks like it's going to be a thoroughly frustrating experience, even worse than initial rollout. For subscription users, the situation is almost indistinguishable from the export ban.

they might as well not released it at all, what's the point of this theater and artificial scarcity

No idea. But I will switch to OpenAI if they release their Sol model on a subscription. And if neither of them do, I will switch to GLM 5.2.

This is what I’m thinking, too. OpenAI is gaining a structural advantage purely on the basis of not being considered an enemy of the administration. Anthropic really blew it with Washington.

By "blew it with Washington" you mean "Didn't donate millions to the ballroom."

It's interesting that the fate of billions or even trillions of dollar hinges on millions of dollars of donations.

That is what corruption usually looks like

Yes. And as the saying goes: the scandal is not that you can buy politicians, the scandal is that they are so cheap.

They blew it by pretending to take some sort of moral high ground while their model was being used in Iran to blow up schools. I say they get what they deserve. I would have a lot more respect for them if they banned Pentagon use of their models outright.

It doesn't look like it; similar restrictions apply to GPT-5.6 as used to apply to Fable.

I think the Fable ban happened because Anthropic was first to release a capable enough model.

i don’t think 5.6 will be as good as fable. their benchmark graphs say so, maybe they’ll take some limiters off next week or something now that being Fable tier isn’t scary anymore.

It will likely be GLM 5.3 by then

Perhaps the 9,999 fields other than computer technician will appreciate it?

So fable will jump more often to Opus than it already did on original release? Working with fable felt like having to constantly fight against your work tool. Frustrating. Now they're making it even more frustrating.

Donald Trump named David Sacks the White House AI and crypto czar. I guess you know whom to thank.

Wasnt it Anthropic marketing their models as very very smart and dangerous?

I can't roll my eyes hard enough at all the people who say this shit about Anthropic every day. I know I'll get downvoted. I know it's lame to complain about future downvotes. I don't care anymore.

Anthropic was correct in their assessment and early warning of Mythos's capabilities, and they did this rollout pretty well. They were not hype marketing. They were being genuinely cautious and honest.

The Trump admin was largely unreasonable with the sudden export control. (Though not entirely unreasonable.) The export control also had not much to do with Anthropic's pre-release warnings. See: GPT-5.6 currently being held up by the federal government.

> Anthropic was correct in their assessment and early warning of Mythos's capabilities, and they did this rollout pretty well. They were not hype marketing. They were being genuinely cautious and honest.

So what prevented them from putting in the sort of safeguards they ended up putting in without hyping it for months prior as being so good, it's too dangerous?

I'm not sure what you're saying. They spent ages adding guardrails to Mythos. Then they spent ages creating a whole new even more guardrailed version of Mythos called Fable. Then they added tons of classifiers so API requests to Fable would get rejected even if you ask a question like "what is a molecule". They put the thickest layer of bubble wrap around the model of any model in history. And then just today they made the classifiers even much more extreme than at the initial launch.

If they were truly honest in their beliefs of the potential risks of this model, how would their behavior have differed? I would expect exactly the behavior we see, if they were being honest in their belief.

Also note Dario here saying they shot themselves in the foot commercially with how they handled the rollout of the model - you can tell by his reflexive reaction how ridiculous he considers the accusation: https://youtu.be/v1wZwxY3CMg?t=2103

I am saying they could have not said anything about it being too dangerous etc. and just released Fable as a new model once the safeguards were in place and Mythos to trusted orgs as they did.

Instead they choose to hype for months about having a model that's simply 'too dangerous to release'.

In other words, why hype it beforehand instead of just quietly add the safeguards they ended up with anyways and release then?

Because Dario is an ASI doomer and deeply fears the power and danger of hyperintelligent AI and it'd be extremely irresponsible to not warn people "hey, this thing can do things the past models couldn't, and the future models may do it even better".

He was trying to be pro-social.

Sacks has been out since March

For reference, here's what my experience with Fable turned out to be like:

https://news.ycombinator.com/item?id=48466313

Just a code review of my own project. Downgraded to Opus 50% of the time while evaluating the critical I/O and memory safety parts, the exact thing I wanted it to do.

And now it's gonna be even worse.

I mean what do you expect when covering memory safety topics with a model that's not allowed to cover security topics? This seems totally expected. It'll be the same when 5.6 is released.

> what do you expect

I expect the strong cybersecurity model to help me strengthen the cybersecurity of my project.

> not allowed to cover security topics

They said it wouldn't be usable for offensive purposes. This is the opposite of that.

You don't have the strong cybersecurity model. That is not Fable. It never was, even at release.

The cybersecurity model is Mythos, which was never made publicly available. It is only available to a list of US government approved companies.

> They said it wouldn't be usable for offensive purposes

No, they said Fable would refuse for cybersecurity and offensive purposes. You are conflating Fable with Mythos.

They're very similar models though, just with different safeguards and restrictions in placae around particular use cases.

I guess the underlying issue is that there is this model that is very capable, but it's being hobbled because of a fear of abuse. It may well be justified, but for a legitimate user any restriction just makes it a worse product and after all the puffery around how good it is (and some practical experience of how good it is) it's a pretty shit experience. "Here's our best model, no you can't really use it".

Fable adds guard rails like cyber refusals to mythos. Mythos is the starting point for fable. Same model family.

Is there a difference though?

Fable 5, harden my openssl project. Then you use the diffs/summary to find out what the bug is for your exploit.

They're going to be verifying people's identities anyway. Why not put that bit of security theater to good use for once? I'm the author of project X, now let the model work on it, would you kindly?

This "only super special corporations get the model" nonsense is dividing society into haves and have-nots.

Fable is very strong for finding bugs. But you are explicitly not supposed to use it for cybersecurity. Even in the initial rollout I had it refuse and fall back to Opus when implementing a change password function

> Fable is very strong for finding bugs.

That's what I was trying to use it for. Find bugs. Anthropic just refused to let it find the memory safety bugs in my C project.

Oh come on Opus is perfectly good enough for any coding task. You will barely notice when it drops down from Fable.

Why use Fable at all then?

Still waiting for the answer.

You probably shouldn't unless you're doing hardcore cybersecurity

It kind of sounds like you aren't allowed to use it for that.

It even refuses to do numerical time-series analysis (this on an empty project). This is something even a non-llm ML algorithm can do. It’s insane

At least subscription users only have to pay $700 for $1000 of extra credits.

> Looks like it's going to be a thoroughly frustrating experience, even worse than initial rollout.

Honestly, why bother with it? They are effectively just releasing the model in-name, but we just get Opus 4.8.

Yeah. I'm gonna ask Fable to code review my other projects and I guess that's it.

It was already frustrating to use before. I wanted to review my own code for OWASP top 10 kind of stuff and it kept refusing. It repeatedly popped up scary warnings about how I was violating TOS. I had to go through quite a few iterations of that prompt. When I finally got it to work it burned through all my remaining usage on a single run.

I won’t even bother with it if they’ve made it even more frustrating. Instead, I’ve been using a combo of Opus 4.8, GLM 5.2 and DeepSeek v4 Pro. Then I have Opus synthesize and verify the reports from all 3 and make the fixes.

“And even though it falls back to Opus, we charge you for Fumble.”

Fable 5 is switching into Opus 4.8 for everything I throw at it. It is not worth it Opus 4.8 is good enough for the next couple months. For $17/mo for pro anthropic is still worth it for claude code etc.

Yes, I am pretty sure it was simply poorly worded.

They almost definitely mean "you will notice even more false positives during seemingly routine coding/debugging tasks than you did at the initial launch". Which is not surprising, given the ordeal they've been put through. Hopefully it won't be too bad.

The main depressing thing for me is it's now only 7 days on the subscription, and then full API pricing, with no mention of even a plan to bring it back to the subscription in the future. (The initial launch mentioned two weeks of subscription, then API pricing, then a hope to return it back to the subscription not long after.)

I wonder if they meant to draw a link between cybersecurity coding and debugging specifically or this really will apply to all coding and debugging. If it really is a more general restriction, then this is practically the same as it still being restricted.

"In the near term" is doing some heavy lifting.

So, you can't use it for coding, can't use it for 'sensitive' information in chemistry/biology. It follows that it's likely bad for medicine and adjacent topics too.

What can you use it for? To run a breadth search on Erdos problems?

To do if/else

In the press release, they 'kind of' clarify this:

   > The new classifier also comes at the cost of flagging benign requests more often during routine coding and debugging tasks.

where did you find that? weird coz their post announcing this also mentioned Claude Code:

> Fable 5 will be available starting tomorrow, Wednesday, July 1, to users globally on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork. For Pro, Max, Team, and select Enterprise plans,1 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. We will re-enable access on AWS, Google Cloud, and Microsoft Foundry as quickly as possible.

https://www.anthropic.com/news/redeploying-fable-5

Their announcement tweet at https://x.com/AnthropicAI/status/2072163884430229756

Reading the full blog post, I think the summary was just poorly written (because it's hard not to read that sentence like all coding is redirected to Opus).

From the full announcement

> The new classifier also comes at the cost of flagging benign requests more often during routine coding and debugging tasks. As with all our safeguards, we’ll continue to refine this to better distinguish genuine misuse from legitimate requests and reduce false positives.

But wasn't the whole (claimed) reason that it got banned in the first place that it is a logical impossibility? Reviewing code for bugs is legitimate. Writing regression tests for bugs is legitimate. If the bug happens to be a security issue then the regression test may be a PoC or at least a step towards one.

Reply on news.ycombinator.com