AI startup Anthropic has officially released its most powerful publicly available model to date.
On June 9, Anthropic announced the launch of its first "Mythos-class" series model for public access, Claude Fable 5. The model is now available to paying users. Concurrently, the latest version of the company's frontier model, Claude Mythos 5, will continue to be offered to a limited number of cybersecurity defenders and infrastructure providers through the Project Glasswing program.
In terms of pricing, both Fable 5 and Mythos 5 are set at $10 per million input tokens and $50 per million output tokens. However, due to anticipated high and unpredictable demand for Fable 5, starting June 23, users will need to spend credits to access it until sufficient compute capacity is restored.
According to the company, Fable 5 surpasses all previously released public models in performance, leading in nearly all AI benchmark tests. It demonstrates exceptional capabilities in software engineering, knowledge work, vision, scientific research, and many other domains. Anthropic confidently stated, "The longer and more complex the task, the greater Fable 5's lead over other models becomes."
Anthropic provided an example citing a usage report from fintech company Stripe, where Fable 5 migrated an entire 50-million-line Ruby codebase in a single day—a task that would take a team over two months to complete manually.
Anthropic also emphasized that Fable 5 is currently the most advanced model for vision tasks. Using vision alone, the model was able to complete the video game "Pokémon: FireRed," whereas earlier Claude models required more complex assistive devices to play the game.
Notably, to ensure safety, Fable 5 will incorporate an "automatic safety downgrade" mechanism, performing automatic safety checks on every user request. If the system detects a request involves one of three high-risk areas (offensive cybersecurity techniques, sensitive biological and life sciences queries, or high-risk jailbreak attempts), it will automatically switch the conversation back to the previous generation's main model, Claude Opus 4.8, for a response.
Dianne Penn, Head of Research and Lab Project Management at Anthropic, stated, "For use cases not involving cybersecurity, we want to release Fable as quickly and safely as possible. Therefore, we decided to release this portion of capability first, while continuing to develop broader cybersecurity applications."
In April, Anthropic revealed it had developed a powerful new AI model called Mythos. Due to potential safety concerns, Mythos was released only in a limited capacity.
Currently, Anthropic is actively advancing towards a public listing alongside its competitor OpenAI. Following its latest funding round, Anthropic's post-money valuation reached $965 billion, while OpenAI's stands at $852 billion, with both companies' potential IPO valuations expected to reach $1 trillion.
On June 1, Anthropic announced on its official website that it had confidentially submitted a draft S-1 registration statement to the U.S. Securities and Exchange Commission (SEC), formally initiating the IPO preparation process. Following the SEC's review, the company can proceed with its listing. According to Anthropic's disclosure, its annualized revenue (ARR) surpassed $470 billion in early May and is expected to reach $500 billion by the end of the month.
On June 8, OpenAI announced it had recently submitted a confidential draft S-1 registration statement. OpenAI indicated that a specific timeline for the public offering has not been set, noting that remaining private may facilitate certain initiatives and that the listing process could take some time. According to media reports citing informed sources, OpenAI also plans to facilitate a tender offer, allowing employees to sell their shares at the latest valuation.
Comments