UnicEdit-10M: A Dataset and Benchmark Breaking the Scale-Quality Barrier via Unified Verification for Reasoning-Enriched Edits

TL;DR

This paper introduces UnicEdit-10M, a large-scale high-quality dataset and UnicBench, a benchmark for reasoning-enriched image editing, addressing the scale-quality trade-off with a unified verification pipeline and a specialized expert model.

Contribution

It presents a scalable data construction pipeline, a 7B expert verification model, and a comprehensive benchmark with new metrics for reasoning in image editing tasks.

Findings

Mainstream models show significant limitations on UnicBench.

The verification pipeline effectively filters high-quality data.

New metrics reveal detailed reasoning capabilities of models.

Abstract

With the rapid advances of powerful multimodal models such as GPT-4o, Nano Banana, and Seedream 4.0 in Image Editing, the performance gap between closed-source and open-source models is widening, primarily due to the scarcity of large-scale, high-quality training data and comprehensive benchmarks capable of diagnosing model weaknesses across diverse editing behaviors. Existing data construction methods face a scale-quality trade-off: human annotations are high-quality but not scalable, while automated pipelines suffer from error propagation and noise. To address this, we introduce a lightweight data pipeline that replaces multi-toolchains with an end-to-end model and a unified post-verification stage. For scalable quality control, we train a 7B dual-task expert model, \textbf{Qwen-Verify}, for efficient failure detection and instruction recaptioning. This pipeline yields \textbf{UnicEdit-10M}, a 10M-scale dataset spanning diverse basic and complex editing tasks. We also propose \textbf{UnicBench}, a general benchmark that extends beyond basic edits to explicitly assess spatial and knowledge-driven reasoning. To enable fine-grained diagnosis, we introduce novel metrics, including \textit{Non-edit Consistency} and \textit{Reasoning Accuracy}. Our analysis of mainstream models on UnicBench reveals their limitations and provides clear directions for future research.