Anthropic Launches Claude Fable 5.1 and Mythos 5.1 With Up to 45% Savings for Agentic Work

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, its latest models for coding and knowledge work. They are the same model with different safeguard levels: Fable 5.1 is generally available, while Mythos 5.1 is restricted to trusted access programs for cybersecurity and life sciences work.



The update lowers the effective cost of running Fable 5.1. Cache reads now cost $0.25 per million tokens, down 75 percent, which Anthropic estimates will cut typical workload costs by about 25 percent compared with Claude Fable 5. Complex coding and highly agentic tasks could see savings of up to about 45 percent. Base pricing remains $10 per million input tokens and $50 per million output tokens.


Safeguards have been adjusted to reduce false positives. Anthropic says its updated cybersecurity protections should result in about 60 percent fewer interventions per session for Claude Code users compared with the previous Fable 5 safeguards. Fable 5.1 can now be used to identify software vulnerabilities for defensive work, though exploit generation and several other dual-use cybersecurity tasks remain restricted.

In biology, the latest safeguards for Fable 5.1 and Fable 5 fire 85 percent less often for benign requests involving elementary biology and medical questions compared with those that launched with Fable 5. Research and development queries in the life sciences will still be directed to Opus models. A new Life Sciences Verification Program, developed with the U.S. government, gives vetted professionals access to Mythos 5.1 with safeguards designed for professional research work.

For enterprise customers, the new Enterprise Frontier Safeguards system keeps data in customer-controlled cloud infrastructure, providing privacy equivalent to a zero-data-retention agreement while still allowing misuse detection. It will roll out in phases beginning this fall, with eligible customers able to use Fable 5.1 under zero data retention until the system becomes available.

Benchmark results show sizable gains over the previous generation. On Terminal-Bench 4.0, Fable 5.1 scores 55.8 percent versus 42.0 percent for Fable 5, while Mythos 5.1 reaches 60.9 percent. Anthropic's published results put Fable 5.1 ahead of Opus 5 and OpenAI's GPT-5.6 Sol across several tests. In a more practical example, investment firm Millennium said Fable 5.1 identified the cause of a rare crash in its internal systems that its engineers and other models had been unable to explain after several years of trying.


The new models were tested on scientific research as well. Mythos 5.1 designed high-affinity protein binders with a hit rate approaching 50 percent across 12 targets, compared with typical rates of 10 to 15 percent in protein design. Anthropic said it was the strongest hit rate it has measured to date.

In another test, Fable 5.1 trained a neural network to create a higher-resolution elevation map covering about a third of Venus from NASA Magellan radar data collected more than 30 years ago. The map resolves details down to two to three kilometers instead of 10 to 20 kilometers and estimates heights up to 25 percent more accurately than before. Anthropic is releasing it under a Creative Commons license.

Fable 5.1 includes invisible text watermarking to meet requirements under the EU AI Act's Code of Practice on Transparency of AI-Generated Content. A detection API is now in private preview for eligible organizations including regulators, media, fact-checkers, researchers, and enterprises with similar compliance obligations.

Claude Fable 5.1 is available today across Anthropic's platforms and cloud partners including AWS, Google Cloud, and Microsoft Azure. Claude Mythos 5.1 is currently available only to a set of U.S. organizations. Anthropic is working with the U.S. government to expand access to a broader group of domestic and international partners.