State of AI Image Generation 2026

AI image generation has reached massive scale, with Adobe Firefly alone past 22 billion assets and Google Gemini the most-used model.

AI image generation is the most widely adopted form of generative media, with personal adoption near 90% and a single tool, Adobe Firefly, surpassing 22 billion generated assets by April 2025. Google's Gemini is now the most-used image model overall, and organizations report fast payback on their generative media investments.

22B

Assets generated by Adobe Firefly by April 2025

Adobe

74%

Adoption of Google Gemini, the most-used AI image model

Artificial Analysis

89%

Strong personal adoption rate for AI image generation

Artificial Analysis

65%

Organizations reporting ROI from generative media within 12 months

Artificial Analysis

Most-Used AI Image Models (Overall Adoption) (%)

Q3 2025 survey; Gemini shown, others ranked

Source: Artificial Analysis

Scale measured in billions

The clearest signal of how mainstream AI imaging has become is raw volume. Adobe announced in April 2025 that Firefly had generated more than 22 billion assets worldwide in under two years. That figure covers only one vendor inside the Creative Cloud ecosystem, so the true cross-platform total is far larger. For buyers, this scale means models have been stress-tested at production volumes that simply did not exist two years earlier.

Image generation outpaces video adoption

Artificial Analysis's Q3 2025 survey found image generation enjoyed strong personal adoption at 89%, well ahead of video. On the organizational side, 44% had image generation running in production versus 39% for video. We attribute the lead to lower cost per output, faster iteration, and clearer use cases like product mockups, marketing creative and concept art. Image generation is effectively the on-ramp through which teams build generative media confidence before tackling video.

Gemini leads, FLUX and OpenAI follow

Google Gemini was the most popular image model in the survey at 74% overall adoption, with OpenAI second and Black Forest Labs' FLUX third. Personal adoption of Gemini (77%) ran ahead of organizational (68%), a reversal of the pattern seen in video. Cost influenced 55% of organizational image-model choices and 46% of personal ones. The takeaway is that capability has commoditized enough that price and integration now drive selection.

Fast payback drives continued spend

Among organizations using generative media broadly, 65% reported a return on investment within twelve months, and 34% described themselves as already profitable on the investment. For image generation specifically, the combination of low per-image cost and high output volume makes the economics easy to justify. We expect spend to keep shifting from experimentation budgets into core marketing and design tooling lines. The constraint going forward is governance and rights clearance, not raw capability.