Archive · 3 Sep 2026
Anthropic Launches Fable 5.1 and Mythos 5.1
Anthropic made two new AI models called Fable 5.1 and Mythos 5.1.
They use the same basic technology but have different safety rules.
Fable is available to everyone, while Mythos is limited to approved organizations.
The models are designed to help with coding, research, cybersecurity, and biology.
Anthropic says Fable 5.1 can be cheaper to use because it charges less when reusing information.
The company says Mythos designed useful protein binders in tests.
Fable also helped create a more detailed map of part of Venus.
Anthropic says its safety systems block fewer harmless requests while still limiting dangerous activities.
The company is restricting Mythos because powerful cybersecurity tools could also help attackers find and exploit software weaknesses.
Anthropic introduced Claude Fable 5.1 for general use and Mythos 5.1 for vetted cybersecurity and life sciences users.
Fable 5.1 workloads are expected to cost about 25% less, or up to 45% less for highly agentic tasks.
Cache-read pricing fell 75% to $0.25 per million tokens, while input and output prices remain $10 and $50.
Anthropic reported stronger coding, research, protein-design, and planetary-science performance for the new models.
The company added more targeted safeguards, citing fewer unnecessary interventions while retaining restrictions on high-risk activities.
- Who
- Anthropic, which developed Claude Fable 5.1 and Claude Mythos 5.1.
- What
- The company launched two AI models with upgraded coding, knowledge-work, research, cybersecurity, and biology capabilities.
- When
- The models were introduced in the reported announcement; Fable 5.1 is immediately available, while Mythos 5.1 access is limited.
- Where
- Fable 5.1 is available through Anthropic's platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure; Mythos 5.1 is offered through Anthropic's verification programs.
- Why
- Anthropic said it aims to improve model capability and affordability while controlling risks from advanced cybersecurity and biology applications.
Broader Access and Capability
Restricted Access and Risk Control
Public availability
Broader Access and Capability
Making Fable 5.1 generally available can give more users access to improved coding, research, and knowledge-work tools.
Restricted Access and Risk Control
Anthropic limits Mythos 5.1 to vetted users because advanced cybersecurity and biology capabilities could create misuse risks.
Cybersecurity use
Broader Access and Capability
Anthropic says the models can identify software vulnerabilities for defensive security work and can reduce unnecessary safety interventions.
Restricted Access and Risk Control
The company continues to restrict activities such as exploit generation and penetration testing because similar capabilities could help attackers.
Scientific claims
Broader Access and Capability
Anthropic reports promising results in protein design and Venus mapping, including a protein-binder hit rate near 50%.
Restricted Access and Risk Control
The reported performance figures are Anthropic's claims, and the articles do not provide independent verification for every benchmark or application.
- General-availability model
- Claude Fable 5.1
- Restricted model
- Claude Mythos 5.1, available only to vetted organizations and professionals
- Typical workload savings
- About 25% compared with Fable 5
- Highly agentic workload savings
- Up to 45% compared with Fable 5
- Cache-read price
- $0.25 per million tokens, a 75% reduction
- Protein-design result
- Nearly 50% of designs were confirmed as viable binders across 12 targets, according to Anthropic
- Venus mapping result
- A higher-resolution elevation map covering about one-third of Venus, with details at roughly 2-3 kilometers compared with 10-20 kilometers previously
- Enterprise Frontier Safeguards
- A system being rolled out that stores customer data on infrastructure controlled by the customer rather than Anthropic's servers
