Anthropic releases Mythos-like model to the public two months after private launch sent shockwaves

In a significant development for artificial intelligence, Anthropic has announced the public release of Claude Fable 5, a model akin to its previously launched Mythos. This decision comes two months after a limited rollout of Mythos, which raised concerns about the model's potential misuse. The company has now lifted the veil on its new offering, aimed primarily at enterprise clients and paid subscribers. The launch of Claude Fable 5 is backed by enhanced safeguards designed to prevent harmful outcomes in sensitive areas such as cybersecurity and biology. Dianne Penn, Anthropic's head of product management for research, emphasized the company's commitment to responsible AI usage. "We focus on a 'race to the top,' providing valuable technology alongside stringent safety measures to ensure that the benefits significantly outweigh any potential harms," she explained in an interview. In April, Anthropic garnered attention from both Wall Street and government sectors with the introduction of Mythos, which specializes in identifying security vulnerabilities in software. Initially, the model was not intended for a broader audience; its rollout was confined to a select group of companies as part of a cybersecurity initiative named Project Glasswing. However, with the unveiling of Claude Fable 5, Anthropic is making strides toward its long-term objective of deploying Mythos-class models widely. Claude Fable 5 demonstrates remarkable performance across software engineering and knowledge-intensive tasks, outperforming its predecessor, Claude Opus 4.8, by over 10% in certain benchmarks. This leap in capability necessitated the implementation of additional protective measures to curb possible misuse. For example, when faced with high-risk inquiries, such as how to create toxic substances, the model will automatically revert to a safer version, Claude Opus 4.8, to provide an appropriate response. On the same day, Anthropic also introduced an updated version of Mythos called Claude Mythos 5, which shares the same core model as Claude Fable 5 but features fewer restrictions in certain contexts. This rollout comes shortly after the company filed a confidential IPO prospectus with the Securities and Exchange Commission, paving the way for what could be a groundbreaking public offering later this year. Anthropic reported that its revenue run rate has surged to $47 billion, a staggering increase from the previous year's $10 billion. The company recently concluded a funding round that valued it at $965 billion, surpassing its main competitor, OpenAI, which was valued at $852 billion earlier this year. OpenAI is also preparing for a major IPO, having recently filed its own prospectus. As competition intensifies, Anthropic CEO Dario Amodei faces the challenge of substantiating the startup's substantial valuation to potential investors. With a pricing model set at $10 per million input tokens and $50 per million output tokens—twice that of Claude Opus 4.8—Claude Fable 5 could prove to be a lucrative asset for the company. Penn highlighted that while pricing is a concern for customers, they are primarily in search of enhanced accuracy and greater value for their investment. "More intelligent models yield a higher return on investment," she asserted.