Short-Story · Short fiction
Lech Mazur · Incorporates required characters, plot, and style naturally into a complete short story.
What it measures and how
Uses 600–800-word stories and 10 creative requirements, with identical prompts, three judges, and comparisons in both directions. Thurstone is primary; Bradley–Terry is diagnostic only. Reasoning tiers follow fixed selection rules.
How this evidence is used
Under review: formal reuse terms and discrepancies between the README and machine results must be resolved before inclusion.
Evaluation limits & attribution
Model judges retain literary preferences; some models have not completed all stories. Short fiction does not establish novel quality or Chinese prose quality.
Data license: No explicit leaderboard-data reuse permission was found. Public GitHub access is not an open license.
Results are published by Lech Mazur. Original scores and the NexGateHub consensus index use different scales and cannot be added directly.