Anthropic has released Claude Fable 5.1 and Mythos 5.1 to address customer complaints regarding pricing, data retention, and excessive safety measures. The company states that Fable 5.1 delivers stronger performance than its predecessor while costing roughly 25 percent less typically. For complex agentic tasks, the price drop reaches up to 45 percent due to reduced charges on cached data that has already been processed and stored. Early testing from Every CEO Dan Shipper suggests the model is the strongest coding tool he has used. He also noted that it is now fast, token-efficient, and speaks like a normal person rather than sounding robotic.
This update matters because it directly targets the cost inefficiencies that have slowed enterprise adoption of agentic workflows. Lowering the price of cached data removes a barrier for teams running repeated inference cycles. The shift towards more natural language output also reduces the friction humans feel when interacting with automated agents. These changes aim to make large language models more viable for practical business applications without sacrificing capability.
- Fable 5.1 costs up to 45 percent less for complex agentic tasks
- Pricing reduction applies to cached data previously processed and stored
- Early users report improved natural language output compared to earlier versions
