Adobe Firefly is stepping up its generative AI game. The company has launched a new set of AI image generation models, a powerful vector design model, and a completely redesigned Firefly web app that brings all of its creative tools under one roof.
This latest version includes Firefly Image Model 4, which now delivers sharper quality, faster rendering, and better control over image style, structure, camera angles, and zoom. It supports resolutions up to 2K. Adobe has also introduced an enhanced version called Image Model 4 Ultra, which can handle detailed scenes with layered structures and small, intricate elements.
According to Alexandru Costin, VP of Generative AI at Adobe, the new Adobe Firefly models were trained with significantly more computing power. This allows for greater image detail, improved text rendering within visuals, and the ability to replicate styles from uploaded reference images.
Firefly Now Powers Image, Video, and Vector Creation
The update doesn’t stop with images. Adobe Firefly now includes its previously beta-only video generation model, which is now available to all users. It enables people to create short video clips using simple text prompts or images. Users can also customize start and end frames, add atmospheric effects, and fine-tune motion design—up to 1080p resolution.
Another big addition is the Firefly Vector Model, built to generate editable vector graphics. This tool lets users create and iterate on logos, product packaging, icons, and even full scenes with just a few clicks. As with its image tools, Adobe Firefly makes it easy to tweak outputs or create variations instantly.
A Unified Web App with Competitive Edge
With the redesigned Adobe Firefly web app, users can now access all of these models in one streamlined space. But here’s the twist—Adobe has integrated models from other top AI labs as well. The app now features OpenAI’s image model, Google’s Imagen 3 and Veo 2, and Flux 1.1 Pro. You can switch between models freely, and every generated image includes content credentials for transparency and attribution.
Adobe also teased more third-party model integrations in the near future, positioning Adobe Firefly as a central hub for creative AI workflows.
Firefly Boards and New Developer Tools
Another new feature in the mix is Firefly Boards, a collaborative canvas for moodboarding and ideation. With Boards, users can generate, remix, or import images and work alongside others—similar to platforms like Kosmik or Visual Electric. Firefly Boards is now part of the web app and offers a more interactive space to plan creative projects.
Developers aren’t left out either. Adobe Firefly’s Text-to-Image API and Avatar API are now generally available, while a Text-to-Video API is open in beta. All of these are part of Firefly Services, a growing collection of APIs and developer tools for bringing generative AI into apps and workflows.
Content Authenticity and Future Integration
Adobe is also testing a separate web app called Adobe Content Authenticity. This tool lets creators attach metadata to their work, confirming ownership and setting usage permissions—including whether AI companies can use that content for training models.
Although Adobe hasn’t announced a timeline, these new Adobe Firefly features will soon integrate into its wider Creative Cloud product suite.
