DALL-E 2 is OpenAI's second-generation text-to-image model, capable of generating, editing, and creating variations of images from text prompts. It introduced inpainting (editing specific image regions via a mask), outpainting (extending images beyond their original borders), and variation generation from an existing image — capabilities that made it useful beyond pure generation.
Outputs are available at up to 1024×1024 resolution. Compared to DALL-E 3, it has weaker prompt adherence and less photorealistic quality, but its editing features and lower cost make it practical for iterative creative workflows that require targeted image modification rather than full regeneration from scratch.
Key Features
Text-to-image generation from natural language descriptions
Inpainting — edit specific masked regions of an existing image
Outpainting — extend images beyond their original canvas boundaries
Variation generation from a source image for creative exploration
Outputs at up to 1024×1024 pixel resolution
REST API access enabling programmatic image creation workflows
Ideal Use Cases
Iterative image editing where only specific regions need modification
Creative asset generation for marketing, social media, or prototyping
Extending product photography backgrounds via outpainting
Generating image variants for A/B testing visual content
Rapid concept art and mood board creation for design teams
Example Prompts for DALL-E 2
Technical Specifications
| Provider | OpenAI |
| Category | Image |
| Modality | Text -> Image |
| Max Output | 1024×1024 px |
Frequently Asked Questions
Try DALL-E 2 now
Start using DALL-E 2 instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.