GPT-Image-2
OpenAI's image generation and editing model, top of the public image leaderboards through mid-2026.
GPT-Image-2 is OpenAI’s image model, used both for generating pictures from a description and for editing existing ones. Through the middle of 2026 it held first place on the public LMArena leaderboards in both the text-to-image and image-editing categories, with rivals from xAI, Google, Meta, Alibaba and ByteDance ranked behind it.
Worth keeping in perspective what that lead measures. Those rankings come from people voting on which of two images they prefer, so they capture immediate visual appeal rather than accuracy, instruction-following, or whether the result is usable in real work. The gaps between the top few models are also small enough that availability, price and editing tools usually matter more in practice than leaderboard position.