Adobe has just released new and improved Firefly AI models along with a redesigned web app, according to TechCrunch. The company, led by Alexandru Costin (Vice President of Generative AI), announced these exciting changes on April 24, 2025. These new tools help people create better images and videos using simple text commands.
What’s New in Adobe’s AI Toolkit
Adobe has launched four main new tools. First, they released Firefly Image Model 4 and Image Model 4 Ultra, which can make better pictures than before. These models can create images up to 2K resolution – that means pictures with lots of detail that look good even when made bigger.
They also introduced the Firefly Vector Model, which creates vector-based artwork. Think of vector art like drawings made of math formulas instead of tiny dots – you can make them bigger or smaller without losing any quality, unlike regular photos that get blurry when enlarged.
The Firefly Video Model, which was in limited testing last year, is now available for everyone. This tool lets you create short video clips just by typing what you want to see. Imagine typing “sunset over ocean waves” and getting a moving video of exactly that!
Finally, Adobe launched Firefly Boards, a collaborative canvas where teams can share ideas and create mood boards together. It’s like a digital bulletin board where everyone can pin their creative ideas.
How These AI Tools Work
Adobe trained these new models with more computational resources – basically, they used more powerful computers to teach the AI how to create better images and videos. They used special high-performance computers called AWS EC2 P5/P5en with NVIDIA Hopper GPUs (Graphics Processing Units – specialized chips that are really good at handling visual information).
The company didn’t share exactly how many GPU hours (the amount of time these special graphics chips spent working) or FLOPS (Floating Point Operations Per Second – a measurement of how many calculations a computer can do each second) they used, but they mentioned it was significantly more than their previous models.
One technical improvement is that the video model’s latency (how long you have to wait for results) was reduced by 60% using NVIDIA TensorRT. This means the system responds much faster when you ask it to create something.
How Adobe Compares to Competitors
Adobe’s new tools compete with other AI companies like OpenAI, Google, and Flux. According to the information provided, Firefly Image Model 4 has better prompt fidelity than OpenAI’s GPT-4o and Google Imagen 3. Prompt fidelity means how accurately the AI follows your instructions – like if you ask for “a red car next to a blue house,” how well does the AI include all those details?
The Firefly 4 Ultra version creates more realistic-looking photos with better details compared to Flux 1.1 Pro. However, OpenAI’s GPT-4o is still better at putting text in images. And Google Imagen 3 offers stronger style options when you want a particular artistic look.
Interestingly, Adobe allows users to switch between these different AI models right in their app. You can try one model for one project and another model for a different task, depending on which one does a better job for what you need.
Key Features and Capabilities
- Content credentials – digital labels that automatically attach to AI-generated work showing which AI model created it (like a digital signature)
- Model switching – ability to choose between Adobe Firefly or other companies’ AI models like OpenAI, Google Imagen 3, Veo 2, and Flux 1.1 Pro
- Transparency – clear information about which AI model was used at each step of creating content
- APIs (Application Programming Interfaces) – connection tools that let other software programs use Adobe’s AI features
- Mobile access – upcoming Firefly apps for phones and tablets (iOS and Android)
Pricing and Availability
| Plan Type | Price | Generative Credits |
|---|---|---|
| Free | ₹0 (Free) | Limited credits for basic features |
| Premium | $199.99 per month (approximately ₹17,034.50) | 50,000 credits for premium features |
Adobe uses a system of generative credits, which work like digital tokens. Every time you create something with AI, you spend some of these credits. Think of them like arcade tokens – each time you play a game (or in this case, create an image or video), you use up some tokens.
All the new Adobe Firefly tools are available now through the Firefly web app. The mobile app for phones has been announced but doesn’t have a specific release date yet.
What This Means for Creators
These new tools make it easier for everyone to create professional-looking images and videos without needing years of experience. For designers in India and around the world, this means faster work and more creative possibilities. The ability to switch between different AI models also gives creators more flexibility to choose the right tool for each specific task.
With the content credentials system, there’s also more transparency about what’s AI-generated, which helps address concerns about fake images. As AI tools become more powerful, Adobe’s approach of combining powerful creation tools with responsible use features could set an important standard for the industry.
