Claude Fable 5.1 succeeds Claude Fable 5 at the top of Anthropic's line-up, at unchanged token prices but with a quarter-price prompt cache. The maker positions it not as the default choice — that is Claude Opus 5, at half the price — but as the model to reach for on demanding reasoning, long-running agentic coding, multistep research and work on documents, spreadsheets and slides. Its distinguishing trait is that adaptive thinking cannot be switched off: every answer is a deliberated one, which is why the maker itself lists its latency as the slowest in the current line-up. On the maker's own figures the generational jump is largest in scientific agent work, where Terminal-Bench-Science climbs from 24.7 to 52.6 percent; the model also reports 73.4 percent on CursorBench 3.2.0 and 65.0 percent on Humanity's Last Exam with tools. Three changes break compatibility with Fable 5: forced tool use now returns an error, earlier models cannot read its thinking blocks, and editing an earlier turn invalidates them. An identical model with tighter safeguards ships as Claude Mythos 5.1, by invitation only.