The last thing FreeCAD needs is AI integration. LLM contributions to the project have been of poor quality so far (just like most LLM contributions to FOSS)
I used to use this & just ended up paying for premium for my family to get over the headache of bugs & risking any IP associated with our google accounts lest Sundar wakes up on the wrong side of the bed
Lol, without any evidence this is just vaporblog, it could just be reudced precision for whatever model either one of them runs & not necessarily a distillation or smaller model to boot or heck even a combo since at this point in time most frontier models are MoEs & getting absurd speeds for 1-20B experts is trivial regardless of batch sizes
reply