Anthropic · LIVEBENCH 2026-06-25

Claude Fable 5 Max Effort

Published benchmark results for this specific model configuration.

claude-fable-5-max-effort

Overall score83.0 / 100

Cost / successful task$1.439USD · benchmark workload

Weight accessNot reported

Where this configuration performs

Category averages from the same release, on a 0–100 scale.

Reasoning89.7

Coding86.0

Agentic Coding62.2

Mathematics96.0

Data Analysis80.5

Language90.7

Instruction Following75.8

Task-level results

theory of mind84.6zebra puzzle100.0spatial96.0logic with navigation78.0code generation91.5code completion80.4javascript68.2typescript53.3python65.0AMPS Hard99.0integrals with game97.0math comp95.1olympiad92.8consecutive events91.4tablejoin56.1tablereformat94.1connections99.3plot unscrambling78.7typos94.0paraphrase77.7simplify72.0story generation71.7summarize81.7

Source files: Scores ↗ Categories ↗ Costs ↗ . Missing cost is unknown, never free. Compare configurations within the same release.