Evaluation sourcesOfficial evaluation
SVGBench · Vector graphic blind selection
SVGBench · Blind-selects SVGs under the same prompt, testing visual clarity and graphic expression.
In NewsObserving
Evidence budgetNot scored
Upstream data as ofTo be confirmed
Last syncedNot collected yet
What it measures, and how
Requires complete blind-test scores that have reached the prompt-coverage threshold; the number of registered models is not the number of valid ranked models. Shares a related task family with other SVG evaluations.
How this evidence is used
Under observation: awaiting stable raw scores, a verifiable update time and the boundaries for re-display.
Limits and data attribution
Web statistics cannot prove that every new model has enough evaluations; without the original table and timestamps, no claim of coverage is made.
Data licence: No clear permission to reuse the scores was found.
Scores published by SVGBench. Raw scores and the News consensus score use different scales and cannot be added directly.