Tech

Claude Fable 5 is back, but I stick to Opus 4.8 for everyday work: 5 reasons why

NurPhoto / Contributor via Getty Images

Follow ZDNET: Add us as a favorite resource on Google.


Highlights taken by ZDNET

  • Legend 5 promises power, but reliability is a concern.
  • A sudden slowdown can derail major AI projects.
  • Opus 4.8 is already perfect for my daily work.

The name “The Legend of 5” is appropriate because, boy, does it have a story to tell. It’s a story of ups and downs, and maybe ups again. It was introduced with great fanfare as a corrupted Mythos, a large language model that can do amazing things but is safe enough for non-government work.

Also: Anthropic’s new Claude Fable 5 is the same base model as the Mythos but with attached guardrails.

But Amazon reported that Fable could not be released. Amazon is said to have the same cybersecurity exploits that Mythos was famous for, but without Mythos’ limited distribution. The legend is that Amazon told the government about this, and the US government responded by slapping export restrictions on Fable.

That was not only to prevent it from being sent outside the US, but also to ensure that no foreigner would be allowed to use it, even if that person was, in fact, within our borders.

Anthropic responded by shutting down access to Fable and Mythos entirely to everyone, regardless of what was done with it, who was using it, or where it was being used. The legend was available for a moment and then went dark. It is still unclear whether Anthropic’s reaction was due to insult or extreme caution.

Now, Fable is back. It is advertised in its app, in the Claude Code terminal, in the Claude app, and just about everywhere. Restrictions on use have been extended for just a few days, until July 7, so people can get a taste of the worst beast ever chained at the behest of America’s security leadership.

But I’m not sure the juice is worth the squeeze, to use an overused but nevertheless incredibly tempting phrase because it’s so well suited to this story.

As I’ve been discussing, the rug was pulled out from under users last month. A legend lowers itself to an inefficient AI when it fears that it might become too powerful or might be used for something too dangerous. I wrote about the turmoil at that time.

Also: Claude Fable 5 secretly fired AI researchers, and the internet went wild

Now I’m seeing reports that Fable is running again, but much more frequently. Reports tell us that whenever there’s even the slightest abuse of security (for example, using the word “security” too soon), the Legend descends into Opus-level performance, as if its new power has simply vanished in a puff of smoke.

So I decided to give Fable 5 a pass, at least for now.

It has long been my practice that, when a new operating system comes out, I don’t install it on my main computer for a few months. This allows all the initial bugs and kinks to work themselves out, allows for fewer updates to be released, and ensures that the worst launch day issues are successfully documented. I think it makes sense to use this practice in Fable 5.

If it stays stable for two or three months (and I know we’re talking AI time here), and the government or Anthropic doesn’t change the rules, again I need something that I’m willing to pay twice the price of the more capable Opus 4.8, and I use Fable. Otherwise, it doesn’t seem that interesting. Here are five reasons why I’m skipping right now.

1. It is very flexible

I progressed beyond using AI to testing AI tools. Now I use Claude Code, Cowork, and OpenAI’s Codex to accomplish real work.

Last weekend, I used a combination of Cowork and Codex to do mission critical mitigation to prevent a spam attack that was quickly turning into a denial of service attack. I worked with two random AIs, until I had a working security solution.

Also: I sent Gemini and Claude to write my email responses – but only one sounded like me

This is not the kind of job where you want your AI model to simply stop working. As I explained about the Codex, I’m on a low level system. I had to get extra rounds out of it a few times by waiting for the timer to expire. It was very frustrating. The difference is that with Codex, all I would have to do is pay more money to keep it going. But if I was relying on Fable, and it got shut down unfairly, it would just go away.

I don’t like it when my tools disappear. So I do my best to use only the most reliable tools. So far, Fable has not earned that label.

2. The guardrails are moving

Based on the Reddit thread, it’s unclear exactly what’s causing the drop from Fable to Opus. I don’t find that acceptable. Of course, we know that AIs are not finite, so they cannot be completely predicted. But it would be good to know, in general, when the project will be shown from the uber model to the inefficient intelligence.

Also: How to stop a massive WordPress spam attack with 4,700 lines of code in two days – thanks to Codex and Claude

So far, Anthropic hasn’t published the exact specifications of the guardrail. My guess is it can’t. It is very responsive to the threat of government bans, so something that may be approved today may be restricted tomorrow.

3. It is very expensive

According to Claude, “Right now, until July 7th, Fable 5 is included in your Max plan (and selected Pro, Team, and Enterprise) at no additional cost, but up to 50% of your weekly usage limits.”

However, Claude said, “After July 7, Fable 5 is completely out of the subscription system. Continued access is moving to a usage bill at standard API rates — $10 per million input tokens and $50 per million output tokens. Basically, that means it stops being part of your Max subscription and becomes a metered add-on.”

Also: AI Model Release Tracker: Anthropic releases Sonnet 5, and Fable 5 is back

I already pay $100 a month for Claude Max and use Opus 4.8. If Fable can go back to what I already have in the subscription, it doesn’t seem too expensive to start using an API charge for an unproven product. I will wait.

4. It may be slow

This is a mixed bag. I see complaints like this one on Reddit reporting that Fable 5 is slow. The Redditor also measured using Ollama on a local server (which I can attest to, is very slow).

On the other hand, another user on X made the case that Fable is not slow, but has a different pacing than Opus.

Also: Anthropic is releasing Claude Tag, its new AI co-founder at Slack

The thing is, whether it’s going slower or moving differently, no one is reporting that the Fable is faster. I spend a lot of time waiting for Claude Code to do his thing. Doubling or tripling its speed may be worth spending more, but staying at the same speed doesn’t provide enough reason to jump ship.

5. Opus 5 is almost here

Both Fable 5 and Mythos 5 are back and available to their respective user bases. Sonnet 5 was released on July 1. That leaves Opus and the baby bear model, Haiku, awaiting the development of “5” versions. Anthropic hasn’t told me anything about Opus 5’s release schedule yet, but you can be sure they’ll have an actual announcement soon.

Also: I tested ChatGPT vs. Claude to see which is better – and if it’s worth changing

Another feature is simple. Opus 4.8 is great. I mean, it’s great. I have been actively using it. It has been doing its job. Yes, today it admitted that it made careless mistakes, but who doesn’t? I’m not sure I need something that is another big expense when the current solution does its job well.

And they all lived happily ever after

So, do I suggest you avoid Fable 5? Not at all. I don’t know what you use AI for, how challenging your projects are, or what your cloud service budget is. All I can tell you is, for my use as a freelance programmer, columnist, and small business owner, I don’t think I need it right now. But that’s me. Be yourself.

Also: Switching to Claude? Here’s how to take your ChatGPT memories with you

When Fable is integrated into the Max system and replaces Opus or is the next level, I will give it a try. I’ve never been against something that requires more creative thinking than Fable has to offer. That’s just for now, of course. Stay tuned. This legend undoubtedly has a long story line. There may be dragons.

Is moving guardrails your dealbreaker when choosing an AI model for heavy duty? Let us know in the comments below.


You can follow my daily project updates on social media. Be sure to sign up for my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtzon Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.



Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button