Srpske NovineSrpske Novine

Anthropic Brings Back Fable 5, Launches Cheaper Sonnet 5

The US lifted export controls on Anthropic's most powerful models, so Fable 5 is available worldwide again, alongside a faster and cheaper new Sonnet 5.

Luka Savić3 min read0
Podeli
Anthropic's Claude artificial intelligence, logo and AI interface on a screen
Anthropic restored Fable 5 and introduced its new Sonnet 5 model./ Foto: Anthropic
In brief
  • The US lifted export controls on Anthropic's most powerful public models on July 1.
  • Fable 5 and Mythos 5 are available worldwide again on the Claude platform, Claude.ai and Claude Code.
  • Anthropic also launched Claude Sonnet 5, with introductory pricing of $2 per million input and $10 per million output tokens through August 31.
  • On Pro, Max and Team plans, Fable 5 is included up to 50 percent of the weekly limit until July 7.

Did an AI tool you used last month suddenly reappear? If you work with Claude models, the answer is probably yes. On July 1, Anthropic re-released Fable 5, its most powerful model that had been frozen over US export controls, and in the same move introduced the newer Sonnet 5.

There is a backstory. Fable 5 first arrived on June 9, then was quickly pulled when Washington imposed export restrictions. The trigger was a finding by an Amazon researcher who showed how the model’s safeguards could be bypassed. Anthropic says it has since added a new classifier that blocks that technique in more than 99 percent of cases. Now that the controls are gone, Fable 5 and the related Mythos 5 are available to users around the world again.

What exactly do you get from today?

Fable 5 is running again on the Claude platform, on Claude.ai, and in Claude Code. For Pro, Max and Team subscribers the model is included up to 50 percent of the weekly limit, and only until July 7. After that date, you spend it through credits. In practice, you get a window of a few days to try the strongest version without an extra bill, then decide for yourself whether pay-as-you-go is worth it later.

Sonnet 5 comes with it, and Anthropic has set it as the default model for Free and Pro accounts. That means you get it the moment you open Claude, with no picking involved. The company markets it as its most agentic Sonnet yet, a model that makes plans, uses tools like browsers and terminals, and can run autonomously. In plain terms, it is aimed at people who write code and build AI agents, not just chat.

Pricing is what will interest many. Through August 31, Sonnet 5 costs $2 per million input tokens and $10 per million output tokens. After that it rises to $3 and $15. Anthropic claims it approaches the performance of stronger Opus models at a noticeably lower cost, which is the classic play, a cheaper model that works almost like the pricier one.

Why does this matter beyond developers?

Because it shows how tied AI models now are to politics, not just technology. A model that worked fine vanished overnight because of a single decision on export controls, and returned just as fast once that decision was reversed. If you build a product or a business around one model, that is a risk you did not have to factor in before.

There is a safety angle too. Anthropic, together with Amazon, Microsoft and Google, proposed a shared industry framework for rating the severity of so-called jailbreak attacks, meaning attempts to break a model’s safeguards. The idea is that everyone measures how dangerous a given flaw really is by the same yardstick, instead of each company judging it on its own.

For an ordinary user the takeaway is simple. You get a stronger model back and a cheaper default option, but also a reminder that access to these tools is not guaranteed. Our advice is to use the window until July 7 and compare Fable 5 and Sonnet 5 on your own real task, because press figures and your actual work are not always the same thing.

And here is the practical question for you, would you rather pay more for the strongest model, or is the cheaper Sonnet 5 perfectly enough for what you do every day?

Podeli
Komentari…
Next storyMicrosoft AI chief: talk of AI consciousness is dangerousMustafa Suleyman, head of Microsoft AI, says talk of possible consciousness in AI models is very dangerous, taking aim at Anthropic and its new Claude Fable 5 model.

From other sections

Komentari (0)

Ostavi komentar

Komentari prolaze kratku moderaciju. Lični napadi i govor mržnje se brišu.

Još nema komentara. Budi prvi i podeli mišljenje.