Krux
April 4, 2026
Adobe Embeds 30+ Outside AI Models Directly Into Creative Apps
Published: April 4, 2026 at 12:38 AM
Updated: April 4, 2026 at 12:38 AM
100-word summary
Adobe quietly published a help page cataloging partner models now baked into Photoshop, Illustrator, Firefly, and Express. You can now generate with OpenAI's GPT Image, Google's Gemini and Veo, Runway Gen-4, or Luma's Ray3 without leaving your canvas. The catch? Each model bills differently. GPT Image costs 60 generative credits per image; Gemini 2.5 Flash costs 10. Video gets pricier: Luma Ray2 at 4K resolution burns 150 credits per second. Adobe promises your prompts won't train anyone's models and embeds Content Credentials so stakeholders know which AI touched each asset. The twist: many models are labeled "limited-time," so that Runway option you just built into your workflow might vanish next...
What happened
Adobe quietly published a help page cataloging partner models now baked into Photoshop, Illustrator, Firefly, and Express. You can now generate with OpenAI's GPT Image, Google's Gemini and Veo, Runway Gen-4, or Luma's Ray3 without leaving your canvas. The catch? Each model bills differently. GPT Image costs 60 generative credits per image; Gemini 2.5 Flash costs 10. Video gets pricier: Luma Ray2 at 4K resolution burns 150 credits per second. Adobe promises your prompts won't train anyone's models and embeds Content Credentials so stakeholders know which AI touched each asset.
Why it matters
The twist: many models are labeled "limited-time," so that Runway option you just built into your workflow might vanish next quarter.