DALL-E
What is DALL-E?
DALL-E is the AI image generation system developed by OpenAI. Originally launched in January 2021, DALL-E has evolved through multiple versions, with DALL-E 3 (October 2023) being the main version and GPT-Image-1 (April 2025) the new native model integrated into GPT-4o.
The name DALL-E is a portmanteau combining artist Salvador Dalí and Pixar's robot WALL-E, reflecting the fusion of art and technology.
Available Models (December 2025)
GPT-Image-1 (April 2025) - Latest Model
- Natively integrated into GPT-4o ("omnimodel")
- Resolution up to 4096×4096 pixels
- Better instruction following than DALL-E 3
- Reliable text rendering in images
- Support for image editing and inpainting
- Conversational multimodal architecture
- Qualities: Low ($0.02), Medium ($0.07), High ($0.19) per image
GPT-Image-1 Mini
- More affordable version of GPT-Image-1
- $0.005-$0.052 per image
- Ideal for high volume and cost-sensitive projects
DALL-E 3 (October 2023)
- Previous model, still widely used
- Better understanding of complex prompts
- Native integration with ChatGPT
- Qualities: Standard ($0.04) and HD ($0.08-$0.12) per image
- Resolutions: 1024×1024, 1024×1792, 1792×1024
DALL-E 2 (Legacy)
- Original model, more affordable
- $0.016-$0.02 per image
- Resolutions: 256×256, 512×512, 1024×1024
- Discontinued for new users
Access and Pricing (December 2025)
Via ChatGPT
| Plan | Price | Image Access |
|---|
| Free | $0 | Limited GPT-4o (~25 images/3h) |
| Plus | $20/month | Full DALL-E 3 + GPT-4o image |
| Pro | $200/month | Unlimited access, priority |
| Team | $25-30/user/month | Shared workspace |
| Enterprise | Custom | SSO, advanced security |
Via API
| Model | Quality | Price/Image |
|---|
| GPT-Image-1 | Low | ~$0.02 |
| GPT-Image-1 | Medium | ~$0.07 |
| GPT-Image-1 | High | ~$0.19 |
| GPT-Image-1 Mini | Low-High | $0.005-$0.052 |
| DALL-E 3 | Standard 1024² | $0.04 |
| DALL-E 3 | HD 1024² | $0.08 |
| DALL-E 3 | HD 1792×1024 | $0.12 |
| DALL-E 2 | 1024×1024 | $0.02 |
Free Alternative Access
- Microsoft Bing Image Creator: Uses DALL-E 3 free with limits
- Microsoft Designer: Integration with DALL-E
- $5 free credits for new OpenAI API users
Main Features
GPT-4o Native Image Generation
- Conversational and iterative generation
- Maintains context between generations
- Editing through natural dialogue
- Character consistency between iterations
- RLHF (Reinforcement Learning from Human Feedback)
DALL-E 3 Capabilities
- Text-to-image from natural language
- Complex prompt understanding without engineering
- Multiple artistic styles
- Improved rendering of hands and faces
- Legible text generation in images
- Variations of existing images
Image Editing (GPT-Image-1)
- Inpainting: modify specific areas
- Outpainting: expand images
- Image-to-image transformation
- Detailed brush control
- Transparent backgrounds (PNG)
Security and Moderation
- Adult/violent content filters
- Public figure protection
- Artist opt-out (style protection)
- C2PA metadata in generated images
- Configurable moderation parameter (auto/low)
Integrations
OpenAI Products
- ChatGPT (Plus, Pro, Team, Enterprise)
- OpenAI API (platform.openai.com)
- Azure OpenAI Service
Third-Party
- Microsoft Bing Image Creator
- Microsoft Designer
- Canva (Magic Studio)
- Adobe Firefly and Express
- GoDaddy Airo
Technical Specifications
| Feature | DALL-E 3 | GPT-Image-1 |
|---|
| Max resolution | 1792×1024 | 4096×4096 |
| Image editing | ❌ | ✅ |
| Inpainting | ❌ | ✅ |
| Streaming | ❌ | ✅ |
| Batch API | ✅ (50% disc.) | ✅ |
| Styles | Natural/Vivid | Prompt-driven |
| Images/request | 1 | 1-10 |
| API Uptime | 99.97% | 99.90% |
Limitations
- Does not generate explicit, violent, or hateful content
- Does not create images of living public figures
- Does not replicate styles of specific living artists
- DALL-E 3: Does not support image editing
- GPT-Image-1: 1 image per request for editing
