Freepik, the web graphic design platform, unveiled a brand new “open” AI picture mannequin on Tuesday that the corporate says was educated solely on commercially licensed, “safe-for-work” pictures.
The mannequin, referred to as F Lite, comprises round 10 billion parameters — parameters being the inner parts that make up the mannequin. F Lite was developed in partnership with AI startup Fal.ai and educated utilizing 64 Nvidia H100 GPUs over the course of two months, in response to Freepik.
F Lite joins a small and rising assortment of generative AI fashions educated on licensed information.
We’ve been secretly engaged on this for months! It feels good to lastly share it!
LINKS:
• Common model: extra predictable and prompt-faithful, however much less creative: https://t.co/MyWsKer9Ir
• Texture model: is extra chaotic and error-prone, however delivers higher textures and… pic.twitter.com/GX5mIpYE8O
— Javi Lopez ⛩️ (@javilopen) April 29, 2025
Generative AI is on the heart of copyright lawsuits in opposition to AI corporations, together with OpenAI and Midjourney. It’s ceaselessly developed utilizing large quantities of content material — together with copyrighted content material — from public sources across the internet. Most corporations growing these fashions argue honest use shields their apply of utilizing copyrighted information for coaching with out compensating the homeowners. Many creators and IP rights holders disagree.
Freepik has made two flavors of F Lite accessible, customary and texture, each of which have been educated on an inside information set of round 80 million pictures. Commonplace is extra predictable and “prompt-faithful,” whereas texture is extra “error-prone” however delivers higher textures and artistic compositions, in response to the corporate.
Right here’s a picture from the usual mannequin generated with the immediate “An individual standing in entrance of a sundown, in majestic environment.”
Freepik makes no declare that F Lite produces pictures superior to main picture turbines like Midjourney’s V7, Black Forest Labs’ Flux household, or others. The objective was to make a mannequin brazenly accessible in order that builders may tailor and enhance it, in response to the corporate.
That being stated, operating F Lite isn’t any straightforward feat. The mannequin requires a GPU with not less than 24GB of VRAM.
Different corporations growing media-generating fashions on licensed information embrace Adobe, Bria, Getty Pictures, Moonvalley, and Shutterstock. Relying on how AI copyright lawsuits shake out, the market may develop exponentially.
