Anthropic, the AI firm behind the Claude language model, has launched its most powerful AI model yet, Claude Fable 5. This is the first publicly available version of its Mythos model, which excels at software engineering, knowledge work, and vision tasks. However, to mitigate potential risks, Anthropic has implemented hard safety limits on high-risk areas such as cybersecurity, biology, chemistry, and distillation.
Fable 5 is a version of the company’s Mythos model, which was initially limited to a handful of partners due to cybersecurity concerns. Last week, Anthropic expanded access to hundreds of organizations across 15 countries, focusing on those that manage critical infrastructure. Now, a version of this technology is available to anyone through Anthropic’s Claude API and consumption-based Enterprise plans.
Access to Fable 5 will be included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost until June 22. After that, it will require usage credits, with plans to restore it as a standard subscription feature as soon as possible. A new version of Mythos, called Mythos 5, is also being deployed to organizations that have already been approved to access the advanced model.
Anthropic’s warning about the dangers of AI comes at a time when the company is preparing to enter the public markets alongside OpenAI and Elon Musk’s SpaceX. The firm has urged major global AI labs to establish a coordinated brake pedal on frontier AI development, citing concerns that systems are advancing so rapidly that they may soon achieve recursive self-improvement (RSI).
To mitigate these risks, Anthropic has stress-tested its classifiers with jailbreak attempts before releasing Fable 5. The company claims that no universal jailbreaks were found in over 1,000 hours of testing, but acknowledges that novel attacks remain possible.
As a result, Anthropic is requiring a 30-day retention on all traffic, even if enterprises previously had zero-retention agreements. This data will not be used for training purposes, but rather to defend against complex and novel attacks, including new jailbreaks, and identify and reduce false positives.
Pricing for Fable 5 and Mythos 5 is $10 per million input tokens and $50 per million output tokens, double the price of Opus 4.8. This may serve as a deterrent for widespread use, given that many enterprises are growing critical of AI costs after seeing the bills come in or blowing through their yearly AI budgets early.
Despite the high price tag, some companies, like shopping rewards platform Rakuten, believe the upside is worth it. “At the highest effort, Fable reflects on and validates its own work,” Rakuten said in a statement. “For us, that’s what makes highly autonomous operations possible — the extra thinking pays for itself.”
Source: Original article