ShareGPT-4o-Image: 91K AI-Generated Images for Multimodal Model Alignment

FreedomIntelligence released a dataset of 91,000 images generated by GPT-4o's image capabilities. The collection includes 45,000 text-to-image and 46,000 text-and-image-to-image samples, designed to align open multimodal models with GPT-4o's visual content creation strengths. The dataset was last updated on July 1, 2025.

Use Cases

• Fine-tuning text-to-image generation models based on the 45,000 text-to-image samples
• Training image editing models based on the 46,000 text-and-image-to-image samples
• Benchmarking open multimodal models against GPT-4o's visual content creation capabilities
• Aligning model outputs with GPT-4o's style and quality in visual content generation

Strengths

• Large-scale collection of 91,000 total image samples
• High-quality images explicitly produced by GPT-4o's generation capabilities
• Includes two distinct task types: 45,000 text-to-image and 46,000 text-and-image-to-image samples

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Data may reflect bias inherent to the GPT-4o model's training and generation process

Provenance

Source

FreedomIntelligence via Hugging Face

Collection Method

Images generated by GPT-4o's image generation capabilities

Freshness

Last updated 2025-07-01 04:36:08; freshness should be verified