An AI model from OpenAI, announced in September 2023, that generates images from text. Its strengths were accurately rendering text inside images — such as “a cafe with a sign that says ‘OPEN’” — and faithfully reflecting the fine details of a prompt in the composition. By being integrated into ChatGPT, it popularized the experience of “making images through conversation.” Its role has now been handed over to the successor GPT Image series, and API access (Images API) ended on May 12, 2026. This article looks at what DALL-E 3 achieved and which options to move to today.
Key Features
- Accurate text rendering inside images: It could draw “text within an image” — signs, logos, poster copy — with high accuracy, an area earlier image generation AI struggled with. This direction has been strengthened further in the successor GPT Image
- High prompt fidelity: It reflected positional and element-level instructions like “a red umbrella on the left, a black cat on the right” in fine detail. Even vague, short prompts were internally expanded into detailed ones, producing images with few compositional breakdowns
- Conversational generation popularized through ChatGPT integration: Without learning any dedicated prompt syntax, users could iterate in a ChatGPT conversation with requests like “make it brighter” or “make it portrait orientation”
- Two quality tiers, standard and HD: Through the API, you could specify standard / hd quality and three sizes — square, landscape, and portrait
- Safety-conscious design: It set out an early policy stance on rights issues in generative AI, such as declining to imitate the styles of living artists and restricting generation of public figures
Pricing
The DALL-E 3 API was retired on May 12, 2026 and can no longer be used for new work. For reference, the API pricing at the time of retirement was as follows.
| Quality / Size | Price per image |
|---|---|
| standard, 1024x1024 | $0.04 |
| standard, 1792x1024 / 1024x1792 | $0.08 |
| HD, 1024x1024 | $0.08 |
| HD, 1792x1024 / 1024x1792 | $0.12 |
To use OpenAI’s image generation today, use ChatGPT (image generation is available even on the free plan, with usage limits) or the successor API models, GPT Image (gpt-image-2 / gpt-image-1 / gpt-image-1-mini).
Pricing is as of August 2026. For the latest pricing, see the official site.
Pros & Cons
✅ Pros
- Set the direction for what is now standard in image generation AI: in-image text rendering and prompt fidelity
- Established an experience anyone could use without prompt-engineering expertise, thanks to ChatGPT integration
- Clear pay-per-image pricing by quality and size made API costs easy to estimate
- Built in rights and safety considerations early, providing reassurance for business use
⚠️ Cons
- The API was retired on May 12, 2026, so new integrations are not possible
- In photorealism and high-resolution output, it lagged behind later and successor models such as Midjourney and GPT Image toward the end
- Partial editing of generated images (inpainting), a feature DALL-E 2 had, was not carried over; the API essentially required regenerating the image
- Maintaining style consistency (the same character in different poses, for example) was a weak point
Comparison with Similar Services
| Comparison | DALL-E 3 | GPT Image | Midjourney | Stable Diffusion |
|---|---|---|---|---|
| Provider | OpenAI | OpenAI | Midjourney | Stability AI |
| Availability | API retired (May 2026) | Available (successor) | Available | Available (open model) |
| Strengths | Text rendering, fidelity | Text rendering, editing, high resolution | Artistic image-making | Local execution, customization |
| Access | — | ChatGPT / API | Web / Discord | Local / various services |
| Free tier | — | ChatGPT free tier | None | Model itself is free |
Who Is It For
- Developers who were using DALL-E 3 via the API and are looking for a migration path (moving to the GPT Image series is OpenAI’s official recommendation)
- Anyone curious where ChatGPT’s image generation came from (today it is handled by the successor, GPT Image)
- Anyone who wants to understand the evolution of image generation AI (text rendering, conversational generation, and safety-first design are elements DALL-E 3 popularized)
Summary
DALL-E 3 raised the standard for image generation AI with its ability to render text accurately inside images and its fidelity to prompts, and through ChatGPT integration it made “creating images through conversation” mainstream. The API was retired on May 12, 2026, and its role has passed to the GPT Image series. If you want OpenAI image generation going forward, choose ChatGPT’s image generation feature or GPT Image via the API — the text rendering and conversational refinement experience DALL-E 3 established is available there at an even higher level.