Anthropic launches Claude Fable 5.1 and Mythos 5.1

Anthropic says Fable 5.1 and Mythos 5.1 are the same model with different safeguards. Fable 5.1 is generally available. Mythos 5.1 is offered only through trusted access programs, with safeguards designed for cybersecurity and life-sciences work.

Benchmarks. Anthropic's figures show Fable 5.1 scoring 52.6% on Terminal-Bench-Science 0.1, up from 24.7% for Fable 5, against 29.0% for Opus 5 and 22.4% for GPT-5.6 Sol. On Terminal-Bench 4.0 it scores 55.8%, and Mythos 5.1 reaches 60.9%. Other listed results include 31.4% on AutomationBench (Fable 5: 17.1%), 73.4% on CursorBench 3.2.0 and 60.9% on Humanity's Last Exam without tools. Anthropic says it ran the evaluations with production safeguards on, which likely lowered some scores.

Price and privacy. Lower cache-read pricing is expected to cut costs by about 25% for typical workloads and by up to about 45% for highly agentic work. A new system called Enterprise Frontier Safeguards stores data in cloud infrastructure that the customer controls. It will roll out in phases starting later this fall. Until then, eligible customers can use zero data retention.

Safeguards. Anthropic says cybersecurity false positives are down 60%. Fable 5.1 can now be used to find software vulnerabilities but not to develop exploits for them. A biology access program for Mythos 5.1, built with the US government, will open to scientists soon.

Science. Anthropic says two outside organizations tested protein binders designed by Mythos 5.1. On three targets, their binding affinities were 10 times higher than the best designs submitted to Adaptyv Bio's competitions. Across 12 targets, nearly 50% of designs were viable binders, compared with a typical 10–15%. Fable 5.1 also used NASA Magellan radar data to train a neural network that produced a higher-resolution elevation map of a third of Venus.

The source text is cut off, so further details were not available.