Anthropic has released Claude Fable 5.2 publicly while reserving Claude Mythos 5.2 for verified organizations working in cybersecurity and biomedical research. The models share the same underlying system, but Fable applies stronger restrictions to sensitive cyber and biology tasks, while Mythos offers approved users access to more advanced capabilities in those areas.
| Model | Access | Cybersecurity and biology capabilities |
|---|---|---|
| Claude Fable 5.2 | Publicly available | Strong filters restrict many sensitive tasks |
| Claude Mythos 5.2 | Approved organizations | Advanced capabilities available through Anthropic’s cybersecurity and life-sciences programs |
Anthropic says the 5.2 generation improves coding, reasoning and visual analysis, with responses up to 30% faster than Fable/Mythos 5.1. Standard API pricing has also fallen to $8 per million input tokens and $40 per million output tokens, down from $10 and $50 respectively for the previous generation. A Fast mode can deliver speeds up to 2.5 times higher, although it costs twice the standard rate.
Internal tests point to longer-running technical work
The company’s most ambitious performance claims come from internal testing and should be viewed accordingly. Anthropic says Fable 5.2 coordinated more than 1,000 sub-agents across eight days to build a functional operating system from scratch, including drivers, storage, windows, login and an installer.
- Anthropic says the models designed three working electronic devices in separate tests.
- The company says they solved two open problems in black-hole physics.
- Anthropic says the system formalized an existing proof of Fermat’s Last Theorem in the Lean proof assistant in about 30 hours.
That Lean result does not mean Claude newly proved Fermat’s Last Theorem; it translated an existing proof into a machine-verifiable formal system. Anthropic also says that, with human guidance, Claude now writes more than 90% of the changes made to the company’s own codebase.
For security work, Anthropic positions Mythos 5.2 as its strongest cybersecurity model so far, saying internal evaluations found more vulnerabilities than Mythos 5.1 and Opus 5.5. The access split continues an approach introduced with Fable 5, whose cyber capabilities were substantially limited before public release.
Fable 5.2 is available through the following services:
- Claude
- Anthropic API
- AWS
- Google Cloud
- Microsoft Azure
Mythos 5.2 remains limited to organizations approved through Anthropic’s cybersecurity and life-sciences programs. The key question is whether the lower pricing and faster throughput will intensify pressure across the AI coding market while the company maintains its stricter gatekeeping for high-risk technical capabilities.
