xAI’s Imagine Image 2.0 lands just behind OpenAI’s GPT-Image-2 in Arena benchmarks

xAI releases Think about Picture 2.0 with modifying instruments and preconfigured templates. The mannequin lands simply behind OpenAI’s GPT-Picture-2 in Enviornment benchmarks.

xAI has launched Think about Picture 2.0 as a brand new “High quality Mode” on grok.com/think about and in Grok’s iOS and Android apps. API entry for Think about Picture 2.0 is coming quickly, based on xAI.

Think about 2.0 is designed to comply with directions with fine-grained accuracy, maintain typography and format clear in complicated visuals, and keep constant throughout a number of generations, the corporate says.

Within the Arena leaderboards as of August 7, 2026, the quicker “low” variant of the mannequin takes second place globally in each classes. It scores an Elo score of 1,439 within the Picture Edit Enviornment, behind OpenAI’s GPT-Picture-2 at 1,463. Within the Textual content-to-Picture Enviornment, it hits 1,320, once more trailing GPT-Picture-2 at 1,380.

xAI

Reve 2.1, Meta’s Muse-Picture, Alibaba’s Qwen-Picture-3.0-Professional, Google’s Gemini, and ByteDance’s SeedDream all rank additional down the checklist. The brand new mannequin additionally beats the “High quality” variant of its predecessor by a large margin in these exams.

Enhancing instruments constructed for iterative workflows

Think about Picture 2.0 ships with a number of modifying options. A instrument referred to as “Magic Wand” modifies solely the chosen space of a picture, based on xAI. A segmentation characteristic lets customers choose exact areas, and a background removing instrument exports topics with a clear background.

 

Multi-Ref Enhancing lets customers mix as much as 5 enter pictures right into a single technology. “Sensible Resize” converts an present picture to any side ratio, with the mannequin filling within the further area by itself.

Templates and video pre-production planning

xAI can also be including templates that bundle widespread picture workflows into preconfigured beginning factors. Classes span picture modifying, product images, advertising and marketing supplies, design instruments, sport belongings, and streaming emojis.

xAI can also be exhibiting a characteristic that generates characters, places, and props individually whereas protecting the visible fashion constant throughout all pictures. The corporate positions this as a stepping stone towards full video manufacturing workflows.

Think about 2.0 can generate constant visible worlds with characters, props, and places that function a place to begin for video manufacturing, xAI says. | Picture: xAI

AI Information With out the Hype – Curated by People

Subscribe to THE DECODER for ad-free studying, a weekly AI e-newsletter, our unique “AI Radar” frontier report six instances a yr, full archive entry, and entry to our remark part.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *