Adobe is leveling up its generative AI game with the release of several new Firefly models and a redesigned web experience, offering creatives more power and precision than ever before. From sharper image generation to fully editable vectors and AI-driven video clips, the Firefly family just got a serious upgrade.
The spotlight is on Firefly Image Model 4, a next-gen image generation tool that delivers higher quality results with better speed and deeper control. Users can now fine-tune outputs with precise adjustments to camera angles, zoom, style, and structure. For those needing even more detail, Adobe is introducing Image Model 4 Ultra, designed to render intricate scenes filled with small elements and rich textures. This model is aimed at users who demand high realism and complexity in their creative assets.
These improvements stem from Adobe’s commitment to investing in serious computing power. Alexandru Costin, VP of Generative AI at Adobe, shared that the models were trained with significantly higher compute capacity. This allows the models not only to generate more refined visuals but also to improve how text is integrated within images. Additionally, users can influence the model’s output by uploading their own images to guide style generation, a key feature for maintaining brand consistency or creative direction.
On the video front, Adobe’s Firefly video model, first introduced in a limited beta, is now widely available. This tool lets users generate video clips directly from text prompts or reference images. Users can define camera angles, specify the beginning and end of shots, and incorporate visual elements like atmospheric effects and motion design—all in resolutions up to 1080p. It’s a clear move towards making high-quality video production more accessible to non-experts.
The new Firefly Vector Model also deserves attention. It’s designed for those working with vector-based formats, making it easy to create and iterate on logos, product designs, icons, and other scalable graphics. This means more flexibility for designers who need editable, layered artwork that remains crisp at any size.
All these tools are now housed in a newly revamped Firefly web app, which acts as a central hub for Adobe’s generative AI tools. The app also includes models from other AI players like OpenAI, Google, and Flux, allowing users to switch between different engines and choose the one that best fits their needs. Every image generated, regardless of the model used, comes with embedded content credentials—metadata that helps maintain transparency and verify authenticity.
In a nod to the collaborative and experimental nature of creative work, Adobe is also rolling out Firefly Boards. This is a digital canvas where users can brainstorm, remix content, and work together in real time. Similar to tools like Visual Electric or Kosmik, Boards is designed for ideation and moodboarding, blending AI-generated content with user-uploaded assets.
For developers and businesses, Adobe is expanding access to its API suite. The Text-to-Image API and Avatar API are now generally available, and a beta version of the Text-to-Video API is being rolled out. These tools are part of the Firefly Services ecosystem, which includes APIs and development tools tailored for embedding AI creativity into apps and workflows.
Looking to the future, Adobe is exploring new ways to support content authenticity. A web app under testing, Adobe Content Authenticity, allows users to embed metadata indicating ownership and usage rights. Creators can also specify whether their content may be used for AI training, adding a layer of control over how their work contributes to future technologies.
As Adobe integrates these tools across its product suite, creatives can expect a smoother, smarter workflow powered by AI. Whether it’s designing a logo, generating a product mockup, or producing a promo video, Adobe’s Firefly platform is becoming a comprehensive toolkit for the modern content creator.
