OpenAI: dall-e-2

DALL-E 2, API id dall-e-2, was OpenAI's image model that takes text input and outputs images. OpenAI's model page states plainly: "DALL·E 2 has been deprecated and removed from the API." The page lists GPT-Image-2.5 Sunburst as the model to use for current image generation and editing. Calls to dall-e-2 should be migrated.

Status

What It Did

• Text to image: text input, image output.
• Inpainting: the only supported feature listed on the model page.
• Endpoints: image generation (v1/images/generations) and image edit (v1/images/edits). Chat, realtime, and audio endpoints were not supported.
• Rate limits (images per minute) on the page: Free 5, Build 2,500, Launch 7,500, Grow 10,000.

The model page does not list pricing or output sizes, so this entry does not either.

Use Cases Today

Because the model is removed, there is no live use case. The entry is useful for:

• Auditing old code for dall-e-2 calls that now fail.
• Understanding which image endpoints a legacy integration relied on before moving to a current model.

Limitation: do not plan new work on this model. Check OpenAI's models documentation for the current image lineup.

Migration Steps

1. Search your code for the dall-e-2 model id.
2. Choose a current image model from OpenAI's models page, starting with the replacement the DALL-E 2 page names.
3. Re-test prompts, sizes, and edit workflows, since a newer model will not reproduce DALL-E 2 outputs.

FAQ

Is DALL-E 2 still available?

No. OpenAI's page says it has been deprecated and removed from the API.

What replaced DALL-E 2?

OpenAI's page recommends GPT-Image-2.5 Sunburst. In this catalog, see GPT Image 2 for a related OpenAI image model.

Could DALL-E 2 edit images?

Yes, inpainting was its listed feature, through the image edit endpoint.

How much did DALL-E 2 cost?

The model page lists no price, so we do not state one.

Alternatives

• GPT Image 2: a newer OpenAI image model in this catalog.
• DALL-E 3: the next DALL-E generation, listed separately here.

Conclusion

DALL-E 2 is a removed model. Treat OpenAI's model page as the source of truth and see the models category for current image options.