Claude Mythos
Model ● Covered in 6 stories + Follow
Claude Mythos is an AI model developed by Anthropic that has been used to discover novel cryptographic weaknesses in encryption algorithms, including a mathematical attack on the HAWK post-quantum signature scheme and improved attacks on reduced-round AES variants. Anthropic researchers demonstrated the model's capabilities in cryptanalysis through iterative prompting, with the research requiring approximately 60 hours and $100,000 in API costs. Anthropic later released Fable, described as a safer variant of the unreleased Mythos model, which outperforms earlier Claude versions on benchmarks and is designed with particular strength in multi-agent tasks.
Updated 6 August 2026
Specifications
No specifications recorded yet.
Latest developments
Discovering cryptographic weaknesses with Claude
Simon Willison's Weblog · 1 month ago ·
28
China's Mythos Moment
ChinaTalk · 1 month ago ·
15
New Claude model - Fable
Ben's Bites · 3 months ago ·
39
Claude Mythos and misguided open-weight fearmongering
Interconnects · 5 months ago ·
31
Q3 2026
Moonshot AI released Kimi K3, an open-weight large language model Model release
- The flaw-hunting machine that isn’t allowed to trust itself
- Anthropic’s Mythos Model Finds Flaws in Strong Encryption Algorithms
- Discovering cryptographic weaknesses with Claude
- China's Mythos Moment
Q2 2026
Anthropic releases Claude Fable 5 model with improved benchmarks and new safety restrictions Model release
Relationships
Products & technology
- Anthropic develops this model · 3 sources
- Integrated with HAWK · 1 source
- Integrated with AES · 1 source
- Integrated with Mozilla Firefox · 1 source
- Anthropic deploys this model · 1 source
Competition
- Competes with Claude Opus 4.6 · 1 source