Second in the world! Musk's xAI releases Grok Imagine Image 2.0 model: AI image generation/editing capabilities greatly enhanced

Elon Musk's xAI has officially released the Grok Imagine Image 2.0 image generation model. This version is now available as a new quality mode on the Grok web interface, iOS and Android mobile apps, with the companion developer API also opened up.

According to the official introduction, Image 2.0 focuses on professional image editing capabilities, making partial modifications and asset reuse its core optimization directions, tailored for real-world scenarios such as commercial design and product photography.

The model's instruction-following ability has been greatly optimized. For complex multi-element compositions, it can autonomously plan layout and typography. The rendering clarity of small text in images has been significantly improved, and the subject settings can be stably preserved during multi-round iterative generation.

The software includes a built-in magic wand local editing tool, allowing users to select a specific area of the image for adjustment without affecting other elements, reducing the workload of repeated drawing.

The segment selection feature supports precise cropping of parts of the image for independent color grading, asset replacement, and other operations. The background removal tool can directly export subject assets with transparent backgrounds, facilitating cross-software recomposition.

It supports inputting up to 5 reference images simultaneously, with the system automatically merging multiple assets, eliminating the manual step of stitching reference images and improving the efficiency of composite image creation.

The smart size adaptation feature covers 9 mainstream aspect ratios from 1:2 to 2:1. When switching aspect ratios, the model automatically fills in missing content without requiring manual edge completion.

xAI has also built multiple preset creation templates into Image 2.0, covering high-frequency commercial scenarios such as product still photography, professional portrait shots, e-commerce promotional images, game concept art, and marketing posters.

The templates come with standardized generation parameters preset, allowing users to quickly produce finished results by simply adding basic assets and text requirements, lowering the barrier to professional image creation.

The latest Arena leaderboard data shows that Grok Imagine Image 2.0 ranks second globally in comprehensive scores across the two tracks of text-to-image and image editing, trailing only the OpenAI GPT Image 2 model.

The model has undergone specialized training for three creative directions: photography, graphic design, and illustration. Iterative optimizations have been made in light and shadow reproduction, material details, and color consistency, reducing issues such as image corruption and element misalignment.