I can tell that these stuff will keep on happening, when deepseek launched, the optimisation they made to make the models cheaper were adopted in Claude and GPTs. These guys are just siphoning off the work the OSS models did & shared for free to everyone.