Just now, Fable 5 has been revived globally! Limited time of 7 days, quota halved.

Just now, Anthropic officially announced: Fable 5 is back!

With that one simple sentence, the whole internet erupted with excitement.

After 19 long days, everyone rushed back to Claude like it was a holiday, just to see that familiar name light up again with their own eyes.

Don't just celebrate—hurry up and use it first!

Because before July 7, even with the top-tier Max and Team plans, the weekly usage limit is hard-capped at half.

And be especially careful: once you hit the cap, the Token consumption for running Fable 5 far exceeds that of Opus 4.8.

Anthropic researcher Alex Albert excitedly said, "Welcome to the Fable world!"

Fable 5 is available for a limited 7 days, with half the quota

First, let's talk about what everyone cares about most: how to use it, how much you can use, and until when.

In the latest blog post "Claude Fable 5 promotional access," Anthropic's official rules are quite straightforward.

There are two ironclad rules as follows——

First, a 7-day countdown.

Starting July 1, ending July 7 at 23:59:59 Pacific Time.

After that, Fable 5 will be removed from subscription plans and can only be used via usage credits on a pay-as-you-go basis.

Referencing API pricing, the input price is $10 per million Tokens, output price is $50 per million Tokens, roughly twice that of Opus 4.8.

Second, 50% weekly usage limit.

Note that Fable 5 only has a 50% weekly usage limit, and other models share the same pool with Fable.

For example: if you've already used half of your quota on other models, the remaining full quota can all be given to Fable 5.

All quota must be used up before July 7, or it expires.

What if you run out? There are two paths:

One is to enable usage credits, billed separately, and continue using Fable 5;

The other is to switch back to other models like Opus 4.8 and continue working within the remaining plan quota.

These 7 days are a public window period. Use it one day, lose it one day.

Finally back, the whole internet celebrates

Honestly, these 19 days have been a roller coaster!

Now, Anthropic has finally put this strongest model back into everyone's hands.

The excitement in the AI world is impossible to hide.

Earlier, a netizen created the website "isfable5back.com," pinging Anthropic's API every minute, with developers around the world refreshing daily.

Today, the page finally jumped from "No" to "YES," and confetti effects instantly burst onto the screen.

The anxiety and waiting of developers have finally turned into ecstasy at this moment.

Most coding tasks are unaffected

When Fable 5 was first released, people found that a simple task would trigger a red line and downgrade to Opus 4.8.

This time, Claude directly announced——

We have updated cybersecurity protections, and the vast majority of coding tasks are unaffected.

However, in the short term, the new protections will slightly over-flag some harmless requests compared to before. Anthropic will continuously optimize in the coming weeks.

When a request is blocked, users will receive a clear notification and the task will be handed over to Opus 4.8.

In effect, Fable 5 has installed a more sensitive "security gate" at its entrance.

The formal name of this security gate is the "safety classifier"——

An independent AI system separate from the main model, specifically designed to detect high-risk requests related to cybersecurity, biochemistry, and distillation.

Token consumption is too fast; some can't take it

Developers who got Fable 5 immediately started testing.

However, developer Ben Davis bluntly said that the model's user experience has been "nerfed epicly."

He complained, "Just ran 3 prompts, each very large, containing many sub-agents. The total API cost for this round is about $90."

His usage quota for 5 hours was still 90% at first, but after just 30 minutes, it plummeted to 30%.

Another developer said that Fable hit its quota limit after just one round of looping, but thankfully the task was completed.

Fortunately, ClaudeDevs officially stated that they have reset the quota for everyone—go ahead and enjoy it.

Tightest security check: Fable 5 rejected your question

As we all know, Fable 5 was partially banned for being "too dangerous," and this return has even more safety measures.

The result is that even more normal requests are being rejected.

For example, Fable5 refused to answer questions related to cancer therapy.

Previously, Professor Derya Unutmaz was quite frustrated about this, and now he challenges Anthropic:

Fable 5 is back, still as useless as before! Hey Anthropic, is this how you support science and care about humanity?

The AI coding benchmark BridgeBench waited 15 days, hoping the strongest AI on earth was back and expecting Fable 5 to dominate the rankings.

As a result, Fable 5 refused to answer in the debugging benchmark, dropping from 9th place last time to 41st place this time.

Still in the deep waters of exponential AI development

On the Remote Labor Index, Fable-5 scored 16.10%.

What is RLI?

The Remote Labor Index uses 240 real remote work projects from professional freelancers, covering 23 fields, involving over $140k worth of human labor.

Each task comes with actual requirements, documents, and accepted human deliverables.

Reviewers then compare AI outputs to human references and judge whether a rational client would accept the result.

That's why the score is still low. Complete projects require planning, document handling, quality control, visual consistency, domain judgment, and final packaging.

Currently, Fable-5 leads the public leaderboard with 16.10%.

AI's ability to automate real remote work is advancing rapidly.

Claude Fable 5 represents a significant jump from the next best model, Opus 4.6's automation rate of 4.2%.

This is a crazy leap.

Earlier, some economists even called Fable 5 a "future Nobel laureate-level PhD student":

If previous AI models were enthusiastic but slightly incompetent PhD students, then Fable 5 is more like some higher-level existence;

Perhaps comparable to a future Nobel laureate-level PhD student.

We are still in the deep waters of exponential development, and even the toughest benchmarks are now being conquered.

7 days, 50% quota—where are you planning to spend it?

Source: Xin Zhi Yuan

Risk Warning and Disclaimer

        Markets are risky; invest with caution. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investing based on this is at your own risk.