The AI image generation race just got a new serious contender. xAI has launched grok-imagine-image-quality, a higher-fidelity variant of its Grok Imagine model, and it has debuted at #5 on both the Artificial Analysis Text-to-Image and Image Editing leaderboards. That makes it the strongest image model outside of OpenAI and Google -- and at a fraction of the price.

Where it sits in the pecking order

The Artificial Analysis leaderboard ranks models using an Elo rating system derived from millions of blind head-to-head votes in their Image Arena -- users compare two images from the same prompt without knowing which model made which. The current top five by Elo are: GPT Image 2 (high) at 1339, GPT Image 1.5 (high) at 1266, Nano Banana 2 (Gemini 3.1 Flash Image Preview) at 1260, Cosmos3-Super-Text2Image (agentic) at 1240, and Nano Banana Pro (Gemini 3 Pro Image) at 1219. grok-imagine-image-quality sits just outside that published top five, making it the leading model from any lab other than OpenAI or Google.

To understand why this matters, you need context on what happened before it. GPT Image 2 launched on April 22, 2026, and immediately hit an Arena Elo of 1,512 -- a 241-point gap over every competitor. The Text-to-Image leaderboard had been one of the most competitive charts in AI for the past year. Google's Nano Banana Pro held the top spot for months, with Flux 2, Grok Imagine, and Reve V1.5 all within striking distance -- the top fifteen models separated by roughly 130 points total. GPT Image 2 didn't just win; it reset the benchmark entirely. grok-imagine-image-quality is xAI's answer to that new bar.