Adobe is repositioning Firefly as a broader creative AI platform, moving it beyond its original role as a standalone image generator. The updated system now covers images, video, audio, and vector graphics, giving creators a wider set of tools inside one ecosystem.
The shift builds on Firefly's existing scale. Since its launch nearly two years ago, Firefly has been used to create over 22 billion assets worldwide, according to Adobe. The new version is designed to extend that usage into more parts of the creative workflow.
Firefly moves beyond image generation
The central change is scope. Firefly is no longer presented only as a tool for generating still images from prompts. Adobe now describes it as a system for producing multiple kinds of digital content, including images, video, audio, and vector graphics.
That matters because many creative projects do not stop at one file type. A campaign, concept pitch, social post, or product mockup may require a still image, a short motion asset, translated audio, and editable graphics. Adobe's update places more of those tasks under the Firefly name.
The expanded platform also keeps prompt-driven creation at the center. Users can describe what they want, refine the result through controls, and move between different content types. The source article frames this as a major rebuild of Firefly rather than a narrow feature update.
Image Model 4 adds Standard and Ultra options
Adobe introduced Firefly Image Model 4 in two versions: Standard and Ultra. Adobe says the Standard model can handle 90 percent of typical creative needs, while Ultra is aimed at more complex, photorealistic scenes.
Both versions are described as improvements over earlier models in rendering people, animals, and architectural structures. Adobe also says text rendering on images has improved, and that the models should follow prompts more accurately.
The text-to-image tool now includes more direct controls for shaping output. According to the source, these include aesthetic filters, the ability to choose specific styles, and options for precise composition adjustments.
For users, the practical implication is clearer iteration. Instead of relying only on a written prompt and repeated retries, creators get more ways to steer the final image. That can make Firefly more useful for work where the layout, style, or visual structure needs to match a specific brief.
Firefly Video Model becomes part of the platform
Adobe has also officially released the Firefly Video Model. It can generate video clips up to five seconds long, with various resolutions up to 1080p and multiple aspect ratios, including 16:9, 9:16, and 1:1.
Adobe presents the model as "commercially safe" because it was trained using licensed Adobe data. Firefly Premium plan members get unlimited access to the Firefly Video Model across all Firefly apps.
The video model includes improvements in photorealism, detail, and the rendering of text, landscapes, and visual effects, according to Adobe. Users can guide a video by uploading start and end frames, or by moving between text, image, and video inputs. Camera movements can also be customized.
That mix of inputs is important for creative control. A user might begin with a written idea, use an image to define a look, and then generate motion from that direction. The source does not describe long-form video creation; the stated limit is clips up to five seconds in length.
New tools cover vectors, boards, audio, and mobile work
The update also adds new Firefly tools around collaboration and production. These include a Text to Vector module for creating editable vector graphics, Firefly Boards for collaborative concept development, and Translate Audio for language translation that preserves the original voice.
Each addition expands Firefly into a different part of the creative process:
- Text to Vector supports editable vector graphics from text prompts.
- Firefly Boards gives teams a space for collaborative concept development.
- Translate Audio handles language translation while preserving the original voice.
Adobe also announced an upcoming app for iOS and Android devices. The mobile version will support image and video creation on the go, with project synchronization across mobile and desktop platforms.
That mobile plan points to a workflow where Firefly projects are not tied to a single device. A creator could start or review a project away from a desktop setup, then continue work later on another platform. The source only states that the app is upcoming and does not provide a release date.
Third-party models arrive with content credentials
Adobe is also opening Firefly to AI models from third-party providers. The company plans to incorporate Google Imagen 3, Veo 2, OpenAI GPT Image Generation, and Flux 1.1 Pro into the Firefly ecosystem.
Future integrations are expected to include models from fal.ai, Runway, Pika, Luma, and Ideogram. This would make Firefly less like a single-model product and more like a front end for choosing among different generation systems.
Adobe says AI-generated content will be labeled with "content credentials" that indicate which model was used. That distinction matters because Adobe stresses that only its proprietary Firefly models are trained exclusively on licensed data and are safe for commercial use.
The source also notes that some third-party model providers, such as OpenAI, are currently involved in legal disputes over their training data. For creative teams, the key takeaway is that Firefly may offer more model choice, but model choice also brings different transparency and commercial-use considerations.
Adobe's Firefly update therefore has two goals at once: broaden what creators can make, and give them clearer information about how AI-generated content was produced. The result is a platform that now reaches across images, video, audio, vectors, collaboration, mobile work, and outside AI models.