Introducing image generation version 2.0 for practical work

Image generation has taken a step toward practical utility with the release of version 2.0. Built with a focus on real work, the system is designed to handle tasks where precision matters, such as creating detailed assets for design projects.

Following complex instructions

A common challenge in visual generation is maintaining clarity when instructions become complex. Image 2.0 follows text prompts closely down to fine details. It plans typography and layout carefully, allowing dense visuals to hold together while keeping small text sharp and legible.

Iterative editing tools

Real-world projects rarely rely on a single generation. To support multi-step workflows, the release introduces targeted editing features:

• A magic wand tool that alters only the selected region.
• Segmentation capabilities to isolate precise areas for modification.
• Background removal that exports subjects with transparent backgrounds.
• Multi-reference editing that accepts up to 5 input images in a single generation.

These additions reduce the need for manual compositing in external software, making it easier to refine assets over multiple iterations. The model is available as the new Quality Mode on grok.com/imagine, through mobile apps, and via API.