Anthropic accompanies the release of Claude Fable 5.2 and Mythos 5.2 with a 20% price reduction and the coordination of 1,000 coding agents that officially completed a proof of Fermat’s Last Theorem in 30 hours instead of 11 days.
First frontier release since the pause call and IPO readiness timing
30% higher speed and 20% price reduction with a 1-million-token context
Frontier Code benchmark leap and 1,000-agent coordination for Fermat proof in 30 hours
Mythos 5.2 restricted to verified organizations in cybersecurity and life sciences
Alignment testing and 1,200% revenue growth in amended prospectus
Frequently asked questions
First frontier release since the pause call and IPO readiness timing
Anthropic released two new frontier artificial intelligence models, Claude Fable 5.2 and Claude Mythos 5.2, on Monday, November 9, 2026, confirming in its official announcement of the new models’ release that they are faster and less expensive than previous versions. The rollout comes as the company prepares for an initial public offering that could take place very soon.
These two models represent the first frontier releases shipped by Anthropic since CEO Dario Amodei’s call in September to slow down the pace of AI development. Regarding this, Diane Penn, head of research product management at Anthropic, stated in an interview with CNBC: “We called for pacing, not a complete stop.”
30% higher speed and 20% price reduction with a 1-million-token context
Anthropic states that the new models generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1, with notable improvements in coding, reasoning, and computer vision tasks according to 9to5Mac reports. Fable 5.2 costs approximately $8 per million input tokens and $40 per million output tokens, according to the official Claude platform developer documentation.
Compared to Fable 5.1’s price of $10 for input and $50 for output, the new pricing represents a 20% decrease according to Yahoo Finance. Both models also feature a 1-million-token context window, a maximum output limit of 128,000 tokens, and a knowledge cutoff date through June 2026.
Frontier Code benchmark leap and 1,000-agent coordination for Fermat proof in 30 hours
Bloomberg reported that Fable 5.2 is capable of handling tougher software engineering work and physical hardware tasks such as robotics control, alongside managing long-running projects and coordinating over 1,000 AI agents simultaneously. The model scored 49.1% on the Frontier Code 2.0 coding benchmark, up from 32.4% for Fable 5.1 according to figures shared by the company with Digital Trends. Additionally, roughly 1,000 Fable 5.2 agents successfully formulated the formal proof of Fermat’s Last Theorem in just 30 hours—a task that took a previous Claude version 11 days, according to Gizmodo.
Diane Penn explained that Fable 5.2 is targeted at “advanced users or super-scale use cases,” noting that the Claude Opus 5.5 model released in September remains a more suitable choice for everyday routine tasks.
Mythos 5.2 restricted to verified organizations in cybersecurity and life sciences
Fable 5.2 is available to all customers via the Claude API, the Claude platform on Amazon Web Services (AWS), and Google Cloud. In contrast, Mythos 5.2 relies on the same base model but with fewer safety restrictions for cybersecurity and life sciences work, making its use strictly limited to organizations certified in Anthropic’s life sciences and cybersecurity verification programs.
According to The Verge, the company first introduced Mythos in April with access restrictions due to its advanced cyber capabilities, and then rolled out the public Fable version two months later with additional safeguards, prior to releasing the 5.1 versions in September.
Alignment testing and 1,200% revenue growth in amended prospectus
Anthropic confirmed that both models underwent rigorous safety and alignment testing alongside the creation of new post-release evaluation and monitoring tools. The company is targeting a valuation between $1 trillion and $1.5 trillion in its public offering, down from the $2 trillion figure discussed in earlier months.
The amended prospectus showed revenues of $33.24 billion for the nine months ended September 30, an increase of nearly 1,200% over the same period the previous year, according to CNBC. Penn stated that the new models handle “entirely novel” work and their capabilities may gradually transition to the Opus category.
Frequently asked questions
Question: How much does Claude Fable 5.2 cost, and how much faster is it than the previous generation?
Answer: It costs $8 per million input tokens and $40 per million output tokens—a 20% decrease from the previous generation—with up to 30% higher generation speed.
Question: What is the difference between Claude Fable 5.2 and Claude Mythos 5.2 in terms of availability and restrictions?
Answer: Both share the same base model, but Fable 5.2 is available to general customers with additional security controls, while Mythos 5.2 is restricted to verified organizations in cybersecurity and life sciences.
Question: What are the most prominent benchmark results achieved by Fable 5.2 in coding and mathematics?
Answer: It scored 49.1% on the Frontier Code 2.0 benchmark compared to 32.4% for the previous version, and roughly 1,000 of its agents completed the formal proof of Fermat’s Last Theorem in 30 hours instead of 11 days.
Question: What was the revenue volume revealed in Anthropic’s amended IPO prospectus?
Answer: The amended prospectus disclosed revenues of $33.24 billion during the 9 months ended September 30, 2026, with annual growth approaching 1,200%.