Adobe unveils three new generative AI models, including the next generation of Firefly

These new models can generate high-quality images, vector graphics, and design templates - all from text.
Written by Sabrina Ortiz, Editor
Adobe Design

Adobe has been quietly sneaking up on the frontrunners in the artifical intelligence (AI) marketplace by developing strong models that millions of creatives have already incorporated into their everyday workflows. Now, Adobe is releasing a slew of models and updates to expand its AI offerings even further. 

On Tuesday, Adobe kicked off its annual Adobe MAX Creativity Conference by announcing its latest innovations, which included three new generative AI models: Firefly Image 2 Model, Firefly Vector Model, and Firefly Design Model. 

Also: Open source isn't ready for generative AI. How stakeholders are changing this light bulb together

In addition, Adobe announced the addition of new AI capabilities across Adobe Creative Cloud apps and Adobe Express. 

Let's look at the announcements in detail, so you can better understand all those new AI offerings. 

Firefly Image 2 Model


Adobe Firefly is the company's take on text-to-image generators, which came out of beta in September, after six months of being in beta. 

Also: As generative AI models evolve, customized test benchmarks and openness are crucial

Since Firefly was first released in March, users have generated over three billion images using the model. Now, Adobe is taking things up a notch by releasing its much more advanced successor, Firefly Image 2. 

The major perk of the new model is the increased quality of renditions, which boast enhanced photographic quality for high-frequency details, higher resolutions, more vivid colors, and improved human renderings, according to Adobe. 

Also: 5 ways to explore the use of generative AI at work

As seen by the comparison images of the first and second models (above), the improved quality of the renditions is noticeable. 

The model will also feature new capabilities, including: Generative Match, which allows users to upload or select an image they'd like the new generation to resemble; Photo Settings, which allows a user to adjust and apply photo settings that resemble those of manual camera lens controls; and Prompt Guidance, which helps users get the results they desire while writing their prompts. 

Firefly Image 2 has also been upgraded to better understand text prompts, ensuring that the image generations are better aligned with the user's vision. 

Also: These are my 5 favorite AI tools for work

Although there are many upgrades, the model will still have all the perks of the first model, including being safe for commercial use, and being trained on Adobe Stock assets and openly licensed, public domain content. 

Firefly Image 2 is available starting today on Firefly.Adobe.com and supports 100-plus languages. 

Firefly Vector Model


Firefly Vector is a brand-new addition to Adobe's generative AI models. It's also the world's first generative AI model for vector graphics, according to the company. 

Vector graphics are often used for business marketing because of their crisp look that can be scaled without compromising the integrity of the image. With the Firefly Vector Model, users will be able to leverage generative AI and use a simple prompt to create "human quality" vectors and pattern outputs. 

Like Firefly Image Model 2, this tool features Generative Match to ensure that the vector outputs match existing styles, such as a brand style kit. The technology is also trained on licensed content, such as Adobe Stock and public domain content where copyright expired. 

Also: Most workers want to use generative AI to advance their careers but don't know how

Other features of the Firefly Vector Model include gradients, which are traditionally tough to complete successfully, organized and user-friendly output, seamless patterns, and precise geometry, according to Adobe. 

The text-to-vector graphic tool is available now in Adobe Illustrator in beta. 

New Firefly Design 


The brand-new Firefly Design model powers a new text-to-template capability, which allows users to use text to generate fully editable templates that meet their exact design needs. 

For example, as seen in the photo above, a user could input text saying they need a flyer for a real-estate open house. The model then generates different design template options that fit the theme. 

Also: Generative AI in commerce: 5 ways industries are changing how they do business

The model can generate all kinds of outputs, including designs for print, social, and online advertising, such as social posts, posters, flyers, digital cards, and more. 

Firefly Design was trained on hundreds of thousands of different Adobe Express templates and also leverages the Firefly Image Model, Adobe Stock, and Adobe Fonts to output the perfect design. 

The text-to-template tool is available in beta in Adobe Express, where the templates are fully editable.  

Adobe Express has also received a wave of new generative AI features, including: Generative Fill, which allows users to easily insert, remove, or replace objects using simple prompts; Translate, which allows users to auto translate into 45 languages; and Drawing and Paint, which adds over 50 multicolor paint and decorative brushes in Express for Education, according to the release. 

Editorial standards