Multimodal Language Models as Text-to-Image Model Evaluators presents MT2IE, a framework where a multimodal large language model generates evaluation prompts and scores images, achieving higher correlation with human judgment than prior metrics. MT2IE recovers official T2I model rankings using only 20 prompts—far fewer than traditional benchmarks—and adapts prompts to each model’s performance, maintaining informative scoring ranges. The approach demonstrates that dynamic, interactive evaluation can replace static benchmarks as T2I models improve.
By Jiahui Chen, Candace Ross, Reyhane Askari-Hemmat, Koustuv Sinha, Melissa Hall, Amy Zhang, Michal Drozdzal, Adriana Romero-Soriano
Imag‑Eval is a new language‑grounded benchmark for evaluating Text‑to‑Image models, focusing on how well they follow compositional natural‑language instructions. It disentangles prompt length from compositional difficulty by independently varying the number of instances and the combination of constraints (rules), providing 1,140 prompts and 8,842 rule combinations. The study shows that for structured skills, the difficulty is mainly driven by the number of grounded rules and their binding to instances rather than prompt length alone.
By Ibrahim Mohamed Serouis, David Jaramillo Duque
ATP‑Bench proposes a new benchmark for evaluating agentic tool planning in multimodal large language models (MLLMs) that generate interleaved text-and-image responses. The benchmark contains 7,702 QA pairs, including 1,592 visual‑question‑answer pairs, across eight categories and 25 visual‑critical intents, all verified by humans. A Multi‑Agent MLLM‑as‑a‑Judge (MAM) system is introduced to assess tool‑call precision, missed opportunities, and overall response quality without relying on ground‑truth references.
By Yinuo Liu, Zi Qian, Heng Zhou, Jiahao Zhang, Yajie Zhang, Zhihang Li, Mengyu Zhou, Erchao Zhao, Xiaoxi Jiang, Guanjun Jiang
The paper introduces QC‑T2I‑Bench, a question‑centric framework that transforms open text‑to‑image prompts into atomic questions and arranges them using Davidsonian Scene Graphs. It employs hierarchy‑constrained aggregation to prune downstream questions when prerequisites fail and to weight simple and complex prompts differently, enabling joint success measurement and comparison of repeated entities across prompts. Evaluations on English and Chinese prompts show that joint completion drops from 80.7% for two‑capability components to 37.2% for seven‑plus components, and the same records are reused for a cost‑aware routing system that achieves ERNIE’s performance with 21.3% fewer GPU‑seconds per million prompts.
By Shaoan Zhao, Fang Zhao, Xueqiang Guo, Xinpei Su, Huanlin Gao, Qiang Hui, Ting Lu, Fuyuan Shi, Chao Tan, Bikun Yang, Kai Wang, Shiguo Lian
arXiv:2606. 31711v1 Announce Type: new Abstract: Faithfulness -- how precisely a generated image aligns with its prompt -- is increasingly central to the real-world utility of text-to-image (T2I) models.
By Yuanhao Ban, Tong Xie, Sohyun An, Yunqi Hong, Evan Frick, I-Hung Hsu, Wei-Lin Chiang, Ion Stoica, Cho-Jui Hsieh
The paper introduces the Generative Embedding Benchmark (GEB), which evaluates how much content from an embedding can be recovered by a decoder that only has access to the frozen embedding and a question, without the original image or intermediate features. GEB uses a curated visual‑question‑answering dataset with 1,800 development and 900 test items covering natural images, scene text, and visual documents. Experiments on seven public embedding models show that visual‑only scores range from 28.25 to 33.21, while joint image‑question encoding boosts scores up to 65.56, revealing that generative readout uncovers information bottlenecks not captured by traditional separability‑based benchmarks.
By Yun Li, Biao Yang, Peixi Wu, Yunhao Zhou, Mingzhou Jiang, Wei Yuan, Fan Yang, Wenwu Ou