OpenAI extends beta access for its DALL-E text-to-image model

OpenAI gives users free credits every month to generate images from DALL-E, along with full usage rights to commercialize the images they create.

OpenAI is giving 1 million people early access to try out the beta version of its advanced text-to-image engine, DALL-E.

When users enter a phrase or series of words into this AI system, it can interpret the description and create multiple images as a result.

Under the beta access, users will receive 50 credits in the first month and 15 free credits every month thereafter. These can each be used to generate one original prompt (creating four images) or an edit or variation of a prompt, creating three images.

Users can also purchase credits in batches of 115 for $15 in the first phase of the beta. Artists in need of financial assistance can also apply for ‘subsidized entry’.

“We are excited to see what people are creating with DALL-E and look forward to hearing user feedback during this beta period,” said OpenAI.

OpenAI said that users who create images from DALL-E will be given full usage rights to commercialize the images, including the right to reprint, resell and resell.

OpenAI made the first version of DALL-E last year. The new version called DALL-E 2 was unveiled in April, which OpenAI said “generates more realistic and accurate images with four times the resolution”.

However, the company chose to keep the text-to-image generator out of public access in order to properly understand the limitations and security concerns of the AI ​​system.

The company said it added new security features as a result, such as refusing uploads of images with realistic faces or attempts to impersonate public figures such as celebrities or politicians.

OpenAI said the system has more precise content filters to prevent harmful images, such as violent, mature or political content. The company has also added a new technique to reduce bias in the AI ​​model, to “more accurately reflect the diversity of the world’s population”.

“Expanding access is an important part of our responsible deployment of AI systems, as it allows us to learn more about real-world use and continue to iterate on our security systems,” said OpenAI.

There is competition for OpenAI as Google Research unveiled its own text-to-image generator called Imagen in May. The Google team behind the model said it had an “unprecedented degree of photorealism” and a deep level of language comprehension.

10 things to know straight to your inbox every weekday. Sign up for the Daily overviewSilicon Republic’s summary of essential sci-tech news.