TextEditBench: Evaluating Reasoning-aware Text Editing Beyond Rendering

TL;DR

TextEditBench is a new benchmark designed to evaluate reasoning-aware text editing in images, focusing on semantic, geometric, and contextual coherence, revealing current models' limitations in complex editing scenarios.

Contribution

We introduce TextEditBench, a comprehensive benchmark with a novel Semantic Expectation dimension to assess reasoning and coherence in text editing within images.

Findings

Current models excel at simple instructions

Models struggle with context-dependent reasoning

Physical and layout consistency remain challenging

Abstract

Text rendering has recently emerged as one of the most challenging frontiers in visual generation, drawing significant attention from large-scale diffusion and multimodal models. However, text editing within images remains largely unexplored, as it requires generating legible characters while preserving semantic, geometric, and contextual coherence. To fill this gap, we introduce TextEditBench, a comprehensive evaluation benchmark that explicitly focuses on text-centric regions in images. Beyond basic pixel manipulations, our benchmark emphasizes reasoning-intensive editing scenarios that require models to understand physical plausibility, linguistic meaning, and cross-modal dependencies. We further propose a novel evaluation dimension, Semantic Expectation (SE), which measures reasoning ability of model to maintain semantic consistency, contextual coherence, and cross-modal alignment during text editing. Extensive experiments on state-of-the-art editing systems reveal that while current models can follow simple textual instructions, they still struggle with context-dependent reasoning, physical consistency, and layout-aware integration. By focusing evaluation on this long-overlooked yet fundamental capability, TextEditBench establishes a new testing ground for advancing text-guided image editing and reasoning in multimodal generation.