Elon Musk's xAI Launches Grok-2 Model, New Model Also Supports Text-to-Image Generation

xAI, an artificial intelligence startup, recently launched Grok-2 and Grok-2 mini. These new models have improved inference capabilities, but are currently in beta rather than official release.

As usual, the Grok-2 series models can be used on the X/Twitter platform, but only for users who have subscribed to X Premium and X Premium +.

In a blog post, xAI stated:

We are excited to release an early preview of Grok-2, a significant upgrade from Grok-1.5, featuring cutting-edge functions for conversational chat, programming development, and reasoning.

We also launched a mini version of Grok-2, which is a smaller-sized version of Grok-2, but its performance is equally powerful.

The early preview of Grok-2 has been tested on the LMSYS leaderboard, under the model name sus-column-r.

xAI plans to offer these two models to developers later this month via enterprise API functionality, meaning developers can use the models by calling an API. However, model usage is not free.

The most noteworthy update with Grok-2 is its ability to generate images from text, and based on current testing, xAI has not set many safety boundaries. Therefore, users can generate a wider variety of images, even those with clearly false content.

Sample images indicate that Grok uses Black Forest Labs' FLUX.1 to rapidly generate images. However, as it is a beta version, xAI has not disclosed much technical detail.

Additionally, Elon Musk plans to deploy the Grok-2 series models on the X/Twitter platform, enhancing X's functionality with artificial intelligence models, including improved search functionality, post analysis, and response features. xAI will also launch a preview version of multimodal understanding in the future.