Claude Fable 5.1 Ships With Cache Reads at a Quarter of the Old Rate

Anthropic released Claude Fable 5.1 on 1 September 2026, alongside Claude Mythos 5.1. Anthropic's model overview describes it as the model for "demanding reasoning and long-horizon agentic work," and recommends it over Claude Opus 5 when evals on Opus 5 at higher effort still fall short.

The specifications carry over from Claude Fable 5: a 1M token context window, 128k maximum output, adaptive thinking always on with a default effort of high, and API pricing of $10 per million input tokens and $50 per million output tokens. The API id is claude-fable-5-1.

What changed is the cache. Across the rest of the current lineup, prompt cache reads cost 10% of the base input price. On Claude Fable 5.1 and Claude Mythos 5.1 they cost 2.5% — $0.25 per million tokens against the $1.00 Fable 5 charged. Anthropic commits to keeping the model available until no sooner than 1 September 2027.

On the Artificial Analysis Intelligence Index v4.1.1, Claude Fable 5.1 scores 66 at its max effort setting, the highest figure on the board; Claude Opus 5 scores 63. Artificial Analysis publishes a separate row for each effort level, so the same model also appears at 65, 62 and 60 for xhigh, high and medium.

Anthropic's documentation now lists Claude Fable 5 under legacy models. It remains available.