xAI's Grok Imagine Image 2.0 climbs to fourth on Artificial Analysis
The benchmark group says the model rose from 18th to 4th in text-to-image, the highest outside OpenAI, at $60 per thousand images against $211 for GPT Image 2.5.
Grok Imagine Image 2.0 took fourth place on the Artificial Analysis Text to Image Leaderboard, the benchmark group said on Friday. That makes it the highest ranked model outside OpenAI. The group described a 14-place climb over the previous generation, from 18th to 4th in text to image and from 16th to 10th in image editing. It placed the model on the Pareto frontier for quality against price.
The model was released in August, according to Artificial Analysis. It supports text to image, image editing and multi-reference generation with up to five input images. The group said the model closed the gap to the leaders on every text-to-image capability and use case it measures. Editing was strongest in animation and gaming, live-action film, and knowledge work.
Posts on Sunday put the Elo score at 1,154 and the price at $60 per thousand images, against about $211 for GPT Image 2.5. The leaderboard is built from votes in the group's Image Arena rather than from a fixed test, so the Elo figure measures preference. The price comparison comes from a follower rather than from Artificial Analysis.
xAI has shipped elsewhere too. Artificial Analysis said on Friday that Grok Voice Transcribe 2.0 took first place on its streaming transcription benchmark, at 2.7 percent word error rate and $0.20 an hour. Followers said they had expected Grok 4.7 instead. That model has not appeared.