Comment by Zephyrion

That's a great question. Right now, the inference costs are coming out of my own pocket. My main goal was to let people experience the model's potential without any barriers.

To keep the project sustainable in the long run, I'm exploring some options, like potentially offering a paid tier for heavy users or more advanced features. For now, I'm focused on improving the core experience and will do my best to keep costs low so it remains accessible to as many people as possible.