Anthropic launches two of its most powerful AI models, with output generation speed increased by 30%.
AI startup Anthropic has released two new-generation AI models, Claude Fable 5.2 and Claude Mythos 5.2.
On Monday, AI startup Anthropic released two new-generation AI models, Claude Fable 5.2 and Claude Mythos 5.2. With the company potentially launching its initial public offering (IPO) as early as next week, the new product launch further demonstrated Anthropic's competitive strength in the frontier AI technology sector.
Anthropic said Fable 5.2 and Mythos 5.2 are its most advanced AI models to date, with output generation speed increased by 30% over the previous generation and usage costs reduced by 20% compared with version 5.1. Both models are priced at US$8 per million input tokens and US$40 per million output tokens.
The launch also marks Anthropic's first new frontier model since CEO Dario Amodei called in September this year for slowing the pace of frontier AI development. Anthropic emphasized that the new models underwent rigorous safety and behavioral alignment testing before release, and the company specifically developed new evaluation mechanisms and post-release monitoring tools to address the risks that advanced AI systems may pose.
"We called for controlling the pace of development, not stopping development," Dianne Penn, Anthropic's head of research product management, said in an interview.
Amid increasingly intense competition in the AI industry, Anthropic needs to strike a balance between technical safety and commercial growth. In its IPO prospectus filed at the end of last month, the company also disclosed in considerable detail the potential risks arising from this tension. As management prepares to launch the IPO roadshow, how to convince investors that its safety-first development strategy can also deliver commercial competitiveness will become a focus of market attention.
According to earlier reports, Anthropic is seeking a valuation range of US$1 trillion to US$1.5 trillion for this IPO, below the US$2 trillion target that had been discussed in the market in recent months. If the listing proceeds smoothly, the deal is expected to become one of the largest IPOs globally.
Anthropic first launched the Mythos model in April this year. Because of its powerful cybersecurity capabilities, the company initially made it available only to a selected group of organizations. Two months later, Anthropic launched the Fable model with additional safety safeguards and made it available to the public.
These two models, especially Mythos, have attracted widespread attention from Wall Street and U.S. government officials due to their advanced technical capabilities, and have also become an important factor driving Anthropic's rapid growth this year.
According to Anthropic's previously updated prospectus, in the nine months ended September 30, the company generated revenue of US$33.24 billion, an increase of about 1,200% from the same period last year, showing continued expansion in enterprise customer demand for advanced AI models.
Regarding the newly released Fable 5.2, Penn said the model is mainly aimed at professional users who need to handle highly difficult tasks, and is particularly adept at complex, open-ended technical work, including fields such as engineering, mathematics and Siasun Robot&Automation technology.
By comparison, Claude Opus 5.5, launched by Anthropic in September this year, is better suited to everyday tasks. Penn noted that when users need to solve problems of high technical difficulty, Fable 5.2 can provide stronger capabilities.
She said these application scenarios involve entirely new frontier technical challenges, and related capabilities may gradually be extended to the Opus series of models in the future, but there are still clear performance differences between the two types of models at present, and frontier applications can benefit from the higher performance of the Fable series.
In terms of product availability, Anthropic said Claude Fable 5.2 was officially launched to all users on Monday, while Claude Mythos 5.2 remains under a restricted release strategy, available only to organizations that have passed review and are participating in the company's life sciences and cybersecurity validation programs.
With improved performance of new models, lower usage costs and the IPO process entering a critical stage, Anthropic is trying to further consolidate its position in the high-end AI model market. However, how to control the safety risks brought by advanced models while expanding commercial applications will remain an important factor in the company's future development and investors' assessment of its valuation.
Related Articles
.png)
Amid a sharp U.S. stock market selloff, the parent company of "GTA 6" bucked the trend and closed higher! First-day sales of US$3.79 billion set a record, and Wall Street's focus shifted to the next-generation GTA Online.

HK Bull/Bear Outstanding Qty Ratio(48:52) | November 21

NVIDIA Corporation (NVDA.US) has plunged 46% over seven weeks, marking its longest losing streak since the 2008 financial crisis, with US$2.6 trillion in market value wiped out.
Amid a sharp U.S. stock market selloff, the parent company of "GTA 6" bucked the trend and closed higher! First-day sales of US$3.79 billion set a record, and Wall Street's focus shifted to the next-generation GTA Online.
.png)
HK Bull/Bear Outstanding Qty Ratio(48:52) | November 21

NVIDIA Corporation (NVDA.US) has plunged 46% over seven weeks, marking its longest losing streak since the 2008 financial crisis, with US$2.6 trillion in market value wiped out.






