Grok Imagine

• 🆕 xAI's image-editing variant within the Grok Imagine family
• 🔧 Edits existing images via natural-language instructions per xAI docs
• 🖼️ Multi-image editing combines up to three source images
• 💬 Drives object replacement, style transfer, and scene composition
• 📅 Released January 2026 in the Grok Imagine lineup
• 🖼️ Part of xAI's broader Grok Imagine visual family
• 🔧 Accessible through xAI's developer Imagine API
• 🏢 Built by xAI

xAI

xAI is an American artificial intelligence company and wholly owned subsidiary of SpaceX. The company develops AI systems under the Grok brand, spanning language models, image generation, and video synthesis. xAI has quickly established itself as a multimodal…

Grok Imagine (edit) is the image-editing member of xAI's Grok Imagine visual-media family. According to xAI's documentation, the Imagine API supports editing existing images via natural-language instructions, with multi-image editing accepting up to three source images in a single request to combine subjects, transfer styles, or compose scenes. It was released in January 2026 as part of the Grok Imagine lineup.

Within the family, it complements the generation-focused Grok Imagine image model and sits upstream of later refinements. The documented editing workflow lets users modify an image through prompts rather than regenerating from scratch, supporting iterative creative changes.

As the earliest edit variant, it was followed by a higher-quality successor, Grok Imagine High Quality (released May 2026), and the SOTA-labeled Grok Imagine High Quality (SOTA) generation model. xAI positions the quality-tier models as offering stronger prompt adherence and improved rendering, so this original edit model serves as the standard, faster baseline for edits.

The broader Grok Imagine lineup also spans video generation, including the Grok Imagine text-to-video and Grok Imagine image-to-video models, reflecting xAI's push beyond its Grok text models into multimodal creative tooling. No independent benchmark figures are available for this specific editing variant, so capability claims here are limited to documented features.