OpenAI is making its famous image generation AI DALL-E available to the public via the DALL-E API. The API is available for public beta access starting November 4 and will make it easier for companies to integrate DALL-E’s image generation capabilities in their products.
Access will be rate-limited, at least in the beginning. The company stated it needs some time to get its systems up to speed and that it won’t be looking into what images the customers generate. Users will also be charged for each individual image they generate with prices spread across three resolution tiers.
- 256 x 256: $0.016/image
- 512 x 512: $0.018/image
- 1024 x 1024: 0.02/image
Interest in text-to-image generation AIs has seen a rise recently, especially over the last year. Microsoft, one of OpenAI’s biggest investors launched the Designer app last month that uses AI to generate imagery for Office users.
Additionally, Shutterstock and Getty Images also announced in October that they’ll be integrating with OpenAI and BRIA respectively to allow users to create custom pictures on their platforms. Since such AI models are trained on mostly stock images, with Shutterstock already licensing its images and relevant metadata to OpenAI in 2021 to train its DALL-E model, royalties will also be paid to any creatives whose generated images end up being used somewhere.
While OpenAI won’t be vetting the images the API’s users generate, it does have some security measures in place in case the AI is used to generate misinformative or harmful imagery. The company has keyword filters in place to prevent things like these from happening, however, how effectively these keywords control image generation is yet to be seen.
This isn’t the first API OpenAI has released either. The DALL-E API joins the likes of GPT-3, Embeddings and Codex in OpenAI’s API platform.
In the News: AMD announce RX 7900 GPUs starting at $899; Sales start December 13