Musk’s xAI has launched Grok image generation model

By Zachary Visconti

Posted on December 9, 2024

Elon Musk’s artificial intelligence (AI) company xAI has launched an image generation model this week, coming just as the Sam Altman-run competitor OpenAI has debuted its highly-anticipated video generation model, dubbed Sora.

On Monday, xAI announced the news in a blog post on its website, highlighting that X users already have access to the tool in select countries, while it plans to roll the feature out to all users within a week. The model can receive inputs in the form of text or other images, and it says it’s able to generate high-quality images in many scenarios where other models have a hard time producing.

The company writes the following in the web post:

We’ve enhanced Grok’s image generation abilities with a new model, code-named Aurora. Aurora is an autoregressive mixture-of-experts network trained to predict the next token from interleaved text and image data. We trained the model on billions of examples from the internet, giving it a deep understanding of the world. As a result, it excels at photorealistic rendering and precisely following text instructions. Beyond text, the model also has native support for multimodal input, allowing it to take inspiration from or directly edit user-provided images.

Despite being launched as “Aurora,” Musk responded to the news saying that xAI is planning to get rid of the name, instead preferring to just use the name Grok that has been used to refer to its text-only model.

“Note: we are deleting the name of the internal @xAI image generation system (Aurora),” Musk wrote in a post on X. “Just think of Grok as doing anything.”

You can see a few examples of the Grok image generation’s pictures below.

Grok was originally launched by xAI last November, offered as a benefit to X Premium Plus users.

The company later open-sourced the model, coming alongside Musk’s criticisms of OpenAI for abandoning its original non-profit mission and not open-sourcing its own model, ChatGPT. In April, xAI also launched Grok 1.5V as its first model with visual processing power.

Currently, xAI is currently upgrading a newly built supercomputing cluster in Memphis, Tennessee, dubbed Colossus, where the company eventually plans to house 200,000 Nvidia H100 and H200 GPUs. Constructed in just 122 days and commission in July, the site is currently home to 100,000 of the GPUs, and Musk said last week that the cluster would eventually include as many as one million of them.

What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send us tips at tips@teslarati.com.