Anthropic is now making its most potent AI model accessible to the general public for the first time, albeit with robust safety protocols in place.
On Tuesday, the artificial intelligence firm introduced Claude Fable 5, the inaugural publicly available iteration of its Mythos model. Anthropic asserts that Fable 5 demonstrates exceptional performance in software engineering, complex knowledge work, and advanced vision tasks, but it operates under strict safety constraints. In highly sensitive domains such as cybersecurity, biology, chemistry, and distillation, the model is programmed to block responses and automatically revert to Claude Opus 4.8.
Mythos, which debuted as a preview in April, was initially restricted to a limited number of partners due to prevailing cybersecurity concerns. Just last week, Anthropic broadened access to hundreds of organizations across 15 countries, maintaining its focus on entities managing critical infrastructure.
Currently, a version of this technology is available to anyone through Anthropic’s Claude API and its consumption-based Enterprise plans. Subscription access will be phased in: until June 22, Fable 5 is included at no additional charge in Pro, Max, Team, and seat-based Enterprise plans. Effective June 23, Anthropic will remove Fable 5 from these plans, requiring usage credits for access thereafter, though the company plans to reinstate it as a standard subscription feature as soon as possible.
Simultaneously, Anthropic is deploying a new iteration of Mythos, named Mythos 5, to organizations that have already secured approval to utilize the advanced model.
The launch of Fable coincides with Anthropic's preparations for its entry into the public markets, placing it alongside industry giants like OpenAI and Elon Musk’s SpaceX. This release also follows the AI firm’s recent plea, urging major global AI laboratories to establish a coordinated "brake pedal" on frontier AI development. Anthropic has cautioned that AI systems are advancing at such a rapid pace that they could soon achieve recursive self-improvement (RSI), meaning they could autonomously enhance themselves without human intervention.
Mindful of the potential capabilities of a Mythos-class model if misused, Anthropic reports that it rigorously stress-tested its classifiers with various "jailbreak" attempts before releasing Fable 5.
“Internally, we ran an external bug bounty that produced no universal jailbreaks in over 1,000 hours of testing. We then worked with external red-teaming orgs which also failed to find universal jailbreaks,” the company stated.
Nevertheless, the possibility of novel attacks remains. Consequently, with the introduction of Fable 5 and Mythos 5, Anthropic will now mandate a 30-day retention period for all traffic, even for enterprises that previously had zero-retention agreements. Anthropic clarified that this data will not be utilized for training purposes, but exclusively to “defend against complex and novel attacks, including new jailbreaks,” and to “identify and reduce false positives.” This policy could establish an industry precedent where access to increasingly powerful models is accompanied by mandatory data retention policies framed as a safety measure.
For those who continue to use the model, it is important to note that not every question will yield a Fable 5 answer. Anthropic indicates that instances where Fable must defer to Opus 4.8 are rare, with early data showing that at least 95% of Fable sessions run entirely on the model’s own responses.
In third-party evaluations, analytics company Hex reported in a statement that Fable was the first model to achieve a 90% score on its core analytics benchmark, which assesses complex, long-running analytical tasks.
“On the hardest questions, it shows strong judgement and attention to nuance,” Hex commented.
Vibe-coding platform Base44 noted in a statement that Fable excels at “one-shotting full apps” and possesses excellent tool-calling capabilities. AI-powered workspace and agent platform Genspark also stated that Fable outperformed every other model in its evaluations, demonstrating significantly better performance on tasks such as UI design and game coding.
The pricing for both Fable 5 and Mythos 5 is set at $10 per million input tokens and $50 per million output tokens, which is double the cost of Opus 4.8. This pricing structure alone may act as a deterrent to widespread adoption.
Many enterprises are becoming increasingly critical of AI costs after experiencing higher-than-expected bills or prematurely exhausting their yearly AI budgets. Advanced models like Opus 4.8 can exacerbate these issues, given their sophisticated reasoning skills that may break down a single request into multiple tasks.
Anthropic anticipates that demand for Fable 5 will be very high and challenging to predict. Indeed, some organizations, such as the shopping rewards platform Rakuten, may deem the potential benefits to be well worth the price point.
“At the highest effort, Fable reflects on and validates its own work,” Rakuten stated. “For us, that’s what makes highly autonomous operations possible — the extra thinking pays for itself.”
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.