• Meta introduced Imagine, a new AI image generation feature.
• The feature uses Meta’s AI model, Llama 3, for sharper, high-quality images.
• The feature is part of a broader announcement Meta made last week.

Meta wants users to let their imagination run wild, with an assist from its AI model.

Meta announced a new image generation feature called Imagine this week that allows users to create images in real time. The company is rolling out the feature in beta to users on WhatsApp and the Meta AI website.

To use it, users simply type the prompt “Imagine” followed by a description of the image they’re trying to generate. Meta said the image will change “with every few letters typed.”

The feature draws upon Meta’s latest open-source large language model, Llama 3, which the company says creates “sharper and higher quality” images and has a “better ability to include text in images.”

Once a user creates an image, they can “animate it, iterate on it in a new style or even turn it into a GIF,” Meta said.

Meta’s image generator is one of several updates to its AI assistant that the company announced last week. The assistant can now accommodate requests as specific as restaurant recommendations with “sunset views” or “vegan options,” offer users interior decor inspiration for a new home, or help them study for a big test by explaining tough concepts.

Meta CEO Mark Zuckerberg referred to the company’s “big AI news” as “pretty wild,” according to Axios. But Meta is still catching up to competitors like OpenAI and Anthropic. The company is still training its larger models of Llama 3, and when released it’s likely to be on par with OpenAI’s GPT-4 and Anthropic’s Claude 3 Opus, which were both released in March.

See Imagine in action below: