Freepik, the web graphics design platform, presented a brand new AI image model on Tuesday, which, in response to the corporate, was only trained on commercially licensed, secure images.
Called the model F liteContains around 10 billion parameters – parameters are the interior components that make up the model. F Lite was developed in collaboration with AI Startup Fal.ai and, in response to FreePik, was trained with 64 NVIDIA H100 GPUs in the middle of two months.
F lite joins a small and growing collection of generative AI models which can be trained on licensed data.
We have been working secretly on it for months! It feels good to finally share it!
Left:
• Regular version: more predictable and faster, but less artistic: https://t.co/Mywsker9ir
• Texture version: is more chaotic and more at risk of errors, but delivers higher textures and … pic.twitter.com/gx5mipye8o
– Report lopez
(@javvilopen) April 29, 2025
Generative AI is at the middle of copyright lawsuits against AI firms, including Openaai and Midjourney. It is commonly developed from public sources on the Internet with massive quantities of content – including copyright -protected content. Most firms that develop these models argue fair use Switches your practice to make use of copyright -protected data for training without compensating the owners. Many creators and IP rights holders don’t agree.
Freepik has made two flavors from F Lite available, standard and texture, each of which were trained on an internal data set of around 80 million pictures. Standard is more predictable and “promptly loyalty”, while the feel in response to the corporate, information from the corporate, provides higher textures and inventive compositions, but provides higher textures and inventive compositions.
Here is an image of the usual model that’s generated with the command prompt “an individual before a sunset in majestic environment”.
Freepik doesn’t claim that f lite creates pictures which can be superior to the leading image generators equivalent to Midjourney's V7, Black Forest Labs' Flux family or others. The goal was to make a model openly available in order that developers could adapt and improve it in response to the corporate.
Apart from that, it isn’t a simple task to run f lite. The model needs a GPU with a minimum of 24 GB VRAM.
Other firms that develop media generator models into licensed data are Adobe, Bria, Getty Images, Moonvalley and Shutterstock. Depending on how Ki -Urhleberrechtslagen, the market could grow exponentially.
