Skip to content
K = 0.731
Benchmark sources

ToneBench · Style and scripts

Towards AI · Writes video scripts in a specified author’s style, assessing openings, structure, expression and spoken pacing.
Official benchmark
Across e/accUnder observation
Evidence budgetNot scored
Upstream data timeUnverified
Last successful syncCollection not started

What it measures and how

10 real script tasks, repeated generations and three judges. Compare only identical tasks, rubrics, judges and run fingerprints. Exclude rows with fallback to other models; do not choose configurations by each column’s highest score.

How this evidence is used

Under observation: data protocol and current coverage are good. Display terms and pure-model configurations still need verification.

Limitations and attribution

Favors the style of one Towards AI channel; exact tasks and reference scripts are not public. The current first-place result includes Fable 5 + Opus 4.8 fallback and cannot represent a single model.

Data license: Public JSON is readable, but explicit data aggregation and display terms have not been found.

Results are published by Towards AI. Raw scores and the e/acc consensus index use different scales and cannot be added directly.