Anthropic drops Claude Fable 5.1 with cheaper pricing and lighter restrictions

Anthropic released two new AI models, Claude Fable 5.1 and Mythos 5.1, and says the update answers customer complaints about price, data handling and overly cautious content filters. The company says Fable 5.1 performs better than its predecessor, Fable 5, while costing roughly 25 percent less for typical use and up to 45 percent less for complex agentic tasks — work where a model plans and carries out multi-step actions with little human input.

Anthropic attributes most of the savings to cheaper pricing on cached data, meaning information the model has already processed and stored so it does not need to process it again on each request.

Early users described the update as a substantial jump from Fable 5. Dan Shipper, chief executive of the software company Every, said in comments Anthropic highlighted: “It’s the strongest coding model we’ve used, but now it’s fast, token-efficient, and crucially actually speaks like a normal person.” Box chief executive Aaron Levie said an agent his company built using Fable 5.1 caught subtleties and ambiguities in a dataset that the same agent running on Fable 5 had missed in an identical test. A commentator posting as Lisan al Gaib pointed to benchmark results showing Mythos 5.1 set to a lower reasoning mode, a setting that trades some accuracy for speed, scoring as well as the previous model set to its highest reasoning mode.

Looser filters, new limits on hacking tools

Anthropic says Fable 5.1 uses “more precise safeguards” that are less likely to block ordinary biology questions than Fable 5 was. Mythos 5.1 keeps the same biology-related restrictions as its predecessor. Anthropic also says it is now allowing Fable 5.1 to identify software vulnerabilities, a use it previously restricted. The company says it will still route some cybersecurity work to its Opus-line models instead, including “penetration testing, exploit generation, and binary-based vulnerability scanning.”

On data handling, Anthropic says a coming feature called Enterprise Frontier Safeguards will offer “complete privacy” by storing customer data on the customer’s own cloud servers rather than Anthropic’s. The company says that feature starts rolling out later this fall, without giving a specific date.

Claude Fable 5.1 is available now on all of Anthropic’s platforms. Mythos 5.1 remains limited to participants in Project Glasswing, a program Anthropic has not detailed publicly. Whether Mythos 5.1 reaches a wider release, and when Anthropic sets a firm date for Enterprise Frontier Safeguards, are the two things to watch next.