LongText-Bench

vision

Multilingual: no

LongText-Bench evaluates text-to-image models on their ability to accurately render long text passages within generated images. It includes English (EN) and Chinese (ZH) subsets to assess multilingual text rendering capabilities.

Methodology

Imported from llm-stats public benchmark metadata. Modality: image. Max score: 1. Categories: image-generation, language, vision. Language: multi. Verified by llm-stats: no.