Imagine 2.0 trails GPT-Image-2 by a hair
xai's imagine 2.0 hits 1439 elo in image edit arena, second only to openai's gpt-image-2. the model introduces advanced editing tools and templates, signaling a shift toward production-ready generative workflows.
xai released imagine image 2.0 as a new 'quality mode' on grok platforms, achieving an elo rating of 1439 in the image edit arena and 1320 in text-to-image, placing just behind openai's gpt-image-2. the model focuses on instruction adherence and consistency, shipping with editing tools such as magic wand for selective modifications and multi-ref editing to blend up to five images. additionally, xai introduced templates for product photography and game assets, alongside features designed to maintain consistent characters and props for video pre-production. api access is forthcoming, positioning xai as a strong competitor in the high-end generative image market.