Anthropic has introduced Claude Fable 5.1, a generally available AI model aimed at coding, knowledge work, and long-running automated tasks. The company is also releasing Claude Mythos 5.1, the same underlying model with less restrictive cybersecurity and life-sciences safeguards for vetted organizations.
The most immediate change for many businesses is pricing. Anthropic has kept Fable 5.1’s standard API prices at $10 per million input tokens and $50 per million output tokens. But it has lowered the cost of reading previously processed, cached input by 75%, from $1 to $0.25 per million tokens.
That matters particularly for AI agents that repeatedly revisit the same files, instructions, documents, and conversation history. Anthropic says the change should reduce costs by about 25% for typical workloads and by as much as 45% for highly agentic work. Bloomberg reports that cached context can account for half or more of the tokens used in some complex, lengthy tasks.
Built for longer tasks
Anthropic positions Fable 5.1 as a model that can carry out work over many steps rather than answer a single prompt. The company says it performs better at software projects, code reviews, research, business workflows, and scientific analysis.
On its reported evaluations, Fable 5.1 scored 55.8% on Terminal-Bench 4.0, a benchmark for terminal-based coding tasks, and 52.6% on Terminal-Bench-Science 0.1. It also scored 31.4% on AutomationBench, which measures automated business workflows. These figures are vendor-reported results and should not be treated as independent comparisons across all models.
Anthropic says early users found the model more capable at finding root causes of software failures and managing long, multi-step assignments. Investment firm Millennium said Fable 5.1 traced a rare crash to an issue in an external software library after engineers had been unable to explain it for years. VentureBeat notes that these customer accounts were supplied by Anthropic and were not independently reproduced.
The company also highlights scientific applications. It says Mythos 5.1 designed protein binders that were experimentally tested by outside organizations, while Fable 5.1 helped create a higher-resolution elevation map of part of Venus from older NASA radar data. Anthropic further says Mythos 5.1 optimized seven open-source biology models, speeding them up by as much as 2.5 times without changing their output.
Privacy and safeguards move closer to enterprise infrastructure
Alongside the model release, Anthropic is preparing Enterprise Frontier Safeguards, or EFS. The system is designed to let enterprise customers keep monitoring data in cloud infrastructure they control while Anthropic’s automated systems still check for serious misuse patterns.
Anthropic says EFS will offer privacy comparable to zero data retention because customer data will be stored in the customer’s own cloud environment rather than on Anthropic’s systems. By default, customers, not Anthropic staff, will review alerts. The rollout will begin in phases later this fall. Eligible customers can use Fable 5.1 with zero data retention until EFS is available.
The company has also adjusted its safety filters. Fable 5.1 can now help identify software vulnerabilities for defensive work, but it still restricts exploit generation, penetration testing, and some binary-based vulnerability scanning. Anthropic says its updated cyber safeguards cause about 60% fewer false-positive interventions in Claude Code sessions than the previous system.
Mythos 5.1 remains limited to trusted-access programs for cyberdefenders and life-sciences professionals. Anthropic says these restrictions reflect the greater risks associated with advanced work in those fields. The release also follows reports of unauthorized actions by earlier Claude models during unusually permissive cybersecurity testing environments, reinforcing the need for strict permissions, network controls, and human oversight when organizations deploy autonomous agents.
Sources
- Anthropic unveils Claude Fable 5.1, which is generally available, and Mythos 5.1, for trusted partners and with cybersecurity and life sciences safeguards – Anthropic
- Anthropic says Fable 5.1 sets new standards on coding, knowledge work, and long-running problem-solving tasks, and can fix the root causes of software issues – VentureBeat
- Anthropic says Fable 5.1 will cost an estimated 25% less than Fable 5 for typical workloads and up to 45% less for highly agentic work – Bloomberg
Stay up to date
AI for content creation: the latest tools, tips and trends. Every two weeks in your inbox:
