Just now, news has excited the entire AI community – Fable 5, which was disabled for many days, appears to be partially restored.
Yesterday afternoon, netizens directly posted solid video evidence!

In the Claude mobile app for a small subset of users, Fable 5 has already prominently appeared in the model selector!


In the Claude Code v2.1.190 update, some users can see Claude Fable 5. It appears in the model selector, or is displayed via the /model command, and after reaching the usage limit, it shows "You've used your included Fable 5 usage for this week."

Multiple netizens have posted screenshots showing the option to select Fable 5.


Anthropic's code in recent days also reveals major information.
On June 24th, according to leaks from multiple bloggers, in the freshly released Claude Code v2.1.190 version update, an extremely eye-catching new line of string was added.
"You've used your included Fable 5 usage for this week."

More crucially, the phrase from the old version, "purchased separately from your plan," was completely removed by the official team.
This sends an extremely strong signal. It's highly likely that Fable 5 will be permanently integrated into the existing Claude subscription service as a "weekly quota"!
This is fantastic news for the vast number of Pro users.
Even more interestingly, this gradual rollout is full of various mysteries.

For example, it's only available in specific conversations.
Some users reported that Fable 5 only takes effect in particular selected conversations; it disappears when switching to a different chat window.

Furthermore, geographical restrictions have also loosened.
A user in Canada excitedly commented: "I successfully sent a few messages! Maybe that rule allowing only Americans has been scrapped?"

Additionally, there was a minor incident.
X user Chetaslua discovered that Fable 5 had reappeared on the AWS (Amazon Web Services) platform, but it wasn't open to everyone.

First hurdle: you must have a US ID and submit it for verification. This time, usage is absolutely tied to real-name authentication.
Second hurdle: submitting a use case and waiting for manual review. After passing identity verification, you also need to submit a specific application scenario description to Anthropic for official manual review.
Only after approval can you enter the whitelist and then obtain actual API call permissions.
Unfortunately, soon after the news broke, it was debunked.
An AWS executive confirmed that Fable 5 on AWS was a UI bug.

Sonnet 5 Early Internal Testing, Performance Crushes the Competition?
Not only that, but foreign influencer @DeepakNesss very calmly dropped another bombshell piece of news:
Claude Sonnet 5 might be landing very soon.


If true, the AI community will be completely transformed once again.
If you are a long-time Claude user, you must know Anthropic's product matrix:
- Haiku: Compact and fast, focusing on cost-effectiveness.
- Sonnet: The backbone, the perfect balance of speed and intelligence, and also the first choice for most daily users.
- Opus: A compute-power guzzler, slower but extremely profound, representing the highest ceiling of model capability.
Currently, we are using models from the 3.5 era, while the rumored Sonnet 5 (Fennec) leaps a whole generation and a half ahead!
Even more terrifying is the phrase "performance close to Opus 4.8."

Source: https://www.cometapi.com/claude-sonnet-5-spotted/
Remember, the current strongest Opus model remains a difficult peak to surpass in logical reasoning, long-context understanding, and complex code refactoring.
If a mid-tier Sonnet model can achieve Opus 4.8 levels, it means "compute power democratization" will truly arrive.
It's highly likely we could obtain deep thinking results in seconds, results that previously required waiting dozens of seconds and incurring high API costs.
And a subtle action by Anthropic's official team further confirms the arrival of the new storm: the existing Claude Sonnet weekly usage limit was just quietly removed by the official team!

In today's world where AI compute power is more expensive than gold, removing the limit for the old Sonnet can only have two reasons:
First, to clear out inventory compute power, making server space for a larger-scale new model.
Second, to have one hundred percent confidence in the stability of the old model, fully unleashing it in preparation for the coronation of the new king.
The Calm Before the Storm,
The "Life-or-Death Exam" in 24 Hours
Why have all these events—code leaks, iOS gradual rollouts, and AWS whitelist verifications for Fable 5—exploded so intensely and concentratedly over the past few days?
It can be said this is a test of limits right on the red line.
Because the Sword of Damocles hanging over Anthropic and the entire US AI industry is about to fall—the June 26th deadline is imminent!
Why is the "June 26th" time node so crucial?
The reason is extreme pressure from Capitol Hill.
Just last week, a bipartisan group of heavyweight US lawmakers unusually sent a joint letter with extremely harsh wording to US Secretary of Commerce Howard Lutnick.

In this letter with a lineup of signatures that can be described as stellar, congressional leaders fired off a series of pointed questions.
1. What specific evaluation criteria does the Commerce Department rely on to decide whether to modify the previous export control ban and ultimately restore public access to this major model?
2. What is the specific timeline for making this life-or-death decision?
At the end of the letter, the lawmakers gave the Commerce Department an ultimatum: A clear answer must be provided by June 26th!
Now, June 26th has arrived. This deadline set by Congress is approaching, and it has brought us the frenzied testing of Fable 5.
Within 24 hours, will we witness an even more massive lifting of restrictions?
Let's hold our breath and wait.
References:
https://x.com/bourneliu66/status/2070053821699743785?s=20
https://x.com/HarshithLucky3/status/2070057959766270430?s=20
https://x.com/chetaslua/status/2070055601108054352?s=20
https://x.com/DeepakNesss/status/2069968386738868229?s=20
This article is from the WeChat public account "New Zhiyuan" (AI Era), author: ASI Apocalypse, editor: Aeneas






