Paper 2509.24900
OpenGPT-4o-Image: A Comprehensive Dataset for Advanced Image Generation and Editing
- Published
- Sep 2025
- Research lab
- Independent
- Citations
- 24
- GitHub
- 33 stars
01 In brief
Summary
OpenGPT-4o-Image is a large-scale dataset for training unified multimodal models in image generation and editing.
It contains 80,000 instruction-image pairs across 11 major domains and 51 subtasks, built using a hierarchical taxonomy and an automated pipeline that leverages GPT-4o.
The taxonomy covers five generation modules (Style Control, Complex Instruction Following, In-Image Text Rendering, Spatial Reasoning, Scientific Imagery) and six editing categories (Subject Manipulation, Text Editing, Complex Instruction Editing, Multi-Turn Editing, Global Editing, Other Challenging Editing).
Fine-tuning leading models on this dataset yields significant performance gains: UniWorld-V1 improves by 18% on ImgEdit-Bench, and Harmon improves by 13% on GenEval.
The dataset outperforms ShareGPT-4o-Image on multiple benchmarks.
The authors note limitations, including potential biases from GPT-4o and reliance on existing benchmarks.
02 From the paper
Abstract
The performance of unified multimodal models for image generation and editing is fundamentally constrained by the quality and comprehensiveness of their training data. While existing datasets have covered basic tasks like style transfer and simple object manipulation, they often lack the systematic structure and challenging scenarios required for real-world applications. To address this bottleneck, we introduce OpenGPT-4o-Image, a large-scale dataset constructed using a novel methodology that combines hierarchical task taxonomy with automated data generation. Our taxonomy not only includes fundamental capabilities such as text rendering and style control but also introduces highly practical yet challenging categories like scientific imagery for chemistry illustrations and complex instruction editing requiring simultaneous execution of multiple operations. Through an automated pipeline leveraging structured resource pools and GPT-4o, we generate 80k high-quality instruction-image pairs with controlled diversity, covering 11 major domains and 51 subtasks. Extensive experiments show that fine-tuning leading models on our dataset achieves significant performance gains across multiple benchmarks, with improvements of up to 18\% on editing tasks (UniWorld-V1 on ImgEdit-Bench) and 13% on generation tasks (Harmon on GenEval). Our work demonstrates that systematic data construction is key to advancing multimodal AI capabilities.