Anthropic Upgrades Claude Fable and Mythos Models With Faster Coding, Lower Costs

The models generate responses up to 30% faster than their predecessors, while their standard application programming interface (API) pricing has been reduced by 20%. Announced on November 9, 2026, the models share the same underlying architecture but differ in their availability and safeguards. Claude Fable 5.2 is available across Anthropic’s platforms, while Mythos 5.2 is restricted to verified organisations working in cybersecurity and biomedical research.

Anthropic said the upgrades bring improvements in coding, reasoning and visual understanding. The company also reported progress in areas such as robotics, hardware engineering and mathematical research during pre-release testing. What are Claude Fable 5.2 and Mythos 5.2?

Claude Fable and Claude Mythos are Anthropic’s models for demanding, open-ended technical work. They are designed to handle complex tasks that require multiple steps, including writing and reviewing code, analysing scientific data and solving engineering problems. While both models use the same underlying system, their access differs.

Fable 5.2 includes cybersecurity and biology safeguards that allow it to be made available to a wider user base. Mythos 5.2, meanwhile, is designed for more advanced cybersecurity and scientific applications, with access limited to organisations that pass Anthropic’s verification programmes. Anthropic said Mythos 5.2 is its most capable model for cybersecurity engineering, with improvements in identifying, reproducing and fixing software vulnerabilities.

On its internal vulnerability-discovery evaluation, Mythos 5.2 found 10.2% more vulnerabilities than Claude Opus 5.5 and 14.4% more than Mythos 5.1. What has changed with the new models? The key improvements include faster response generation, lower API costs and stronger performance across technical tasks.

Anthropic said the models generate output up to 30% faster than their respective 5.1 versions. Standard API pricing has also dropped from $10 per million input tokens and $50 per million output tokens to $8 and $40, respectively. An optional fast mode can increase token-generation speed by up to 2.5 times, at a higher price of $16 per million input tokens and $80 per million output tokens.

The company also highlighted several applications of Fable 5.2 during testing. In one experiment, the model coordinated more than 1,000 AI agents to build a prototype operating system over eight days. The system included a kernel, hardware drivers, file storage, a login screen and an installer.

Anthropic also said a group of 1,000 Fable 5.2 agents formalised a proof of Fermat’s Last Theorem in 30 hours, a process that involves translating mathematical reasoning into a form that can be verified by computers. The company reported that Fable 5.2 solved two previously open problems in theoretical physics related to black holes. These results are based on Anthropic’s internal testing and research.

The company said the models are intended for complex technical work, while Claude Opus 5.5 remains its recommended model for most everyday coding and knowledge tasks. The release reflects Anthropic’s push to develop AI systems that can work across longer, more complex tasks, while improving the cost and speed of deploying them.