TubeLab · Video scripts
TubeLab · Writes scripts for different video channels, comparing story organization, style, pacing and clarity.
Across e/accUnder observation
Evidence budgetNot scored
Upstream data timeUnverified
Last successful syncCollection not started
What it measures and how
A common TubeLab Scriptwriter, 12 tasks and 6 writing-quality criteria. Three model judges do not judge their own organization’s models; failed tasks retain penalties. Cost and speed do not affect capability scores.
How this evidence is used
Under observation: stable raw data, accurate evaluation dates and reuse terms are pending. Display pages do not replace verifiable results.
Limitations and attribution
The Scriptwriter workflow affects models. OpenAI uses the lowest reasoning tier, so results cannot be generalized to other tiers. Spoken video scripts are not screenplays.
Data license: Stable result exports and public aggregation permission have not been confirmed.
Results are published by TubeLab. Raw scores and the e/acc consensus index use different scales and cannot be added directly.