In a major AI update, Adobe launches Firefly Image Model 4, a powerful upgrade to its generative image AI, along with new tools for video and vector generation, a fully redesigned Firefly web app, and expanded cross-model access from platforms like OpenAI, Google, and Flux.
This overhaul reflects Adobe’s increasing commitment to AI-first creative workflows, and positions the company as a serious competitor in the rapidly growing creative AI landscape.
Firefly Image Model 4 & Image Model 4 Ultra
Adobe’s Firefly Image Model 4 delivers significant improvements in image quality, realism, control, and resolution — now supporting outputs up to 2K. Users can fine-tune structure, camera angles, zoom, and even match the model’s style to reference images.
A more advanced variant, Image Model 4 Ultra, is designed to handle complex compositions, excelling at rendering intricate structures and fine visual details.
“Compared to previous versions, these models are trained with far more compute power, allowing for sharper text in images and richer detail,” said Alexandru Costin, VP of Generative AI at Adobe.
Video and Vector Generation Comes to Firefly
The long-awaited Firefly Video Model — first seen in beta last year — is now open to all users. This tool lets creators generate high-quality 1080p video clips from text prompts or reference images, with features like:
- Start and end frame control
- Camera angle adjustment
- Motion design customization
- Atmospheric effects generation
In parallel, Adobe introduced the Firefly Vector Model, designed to generate editable vector-based artwork such as logos, icons, packaging, patterns, and product mockups — ideal for branding and design professionals.
Redesigned Firefly Web App: One Platform, Multiple Models
Adobe has also completely redesigned the Firefly web app, centralizing all its generative models in one place. And it’s not limited to Adobe’s models — the platform now includes:
- OpenAI’s image models
- Google’s Imagen 3 and Veo 2
- Flux’s 1.1 Pro model
Users can seamlessly switch between these models and compare results. All generated images, regardless of model, will include content credentials to ensure traceability and authorship.
New Feature: Firefly Boards for Moodboarding
Adobe also unveiled Firefly Boards, a collaborative ideation space where users can generate, remix, and organize visuals into moodboards. Similar to apps like Visual Electric or Kosmik, it aims to speed up creative brainstorming and client presentations. Boards is available now in the Firefly web app.
APIs and Content Attribution Tools
To support developer integration, Adobe has released several Firefly APIs including:
- Text-to-Image API (general availability)
- Avatar API (general availability)
- Text-to-Video API (beta)
All are part of the Firefly Services collection, designed to help developers embed generative AI into third-party apps and workflows.
In addition, Adobe is continuing to promote digital content credentials through its Adobe Content Authenticity web app, which allows users to embed attribution metadata and flag whether an image can be used for AI model training. This tool is being beta tested alongside Adobe’s broader efforts to build a transparent ecosystem for AI-generated content.
The Road Ahead
While Adobe hasn’t confirmed exactly when these new tools will be fully integrated across Creative Cloud apps like Photoshop and Illustrator, the company assures that this is the next step.
As AI-generated content becomes more mainstream, Adobe’s Firefly updates show a clear direction: multi-modal creation, model transparency, and creative control — all wrapped into one accessible platform.
Get the Latest AI News on AI Content Minds Blog