Because find me a person who thinks 25% of their income is “next to nothing”

Meanwhile, a tax avoiding billionaire will double their net worth and take out an asset-backed loan of 100mil to live off and then pay ~0% of that in taxes.

For me, i don’t personally know anybody with enough vram to be running more than a 4bit quant - so that’s my line.

As far as i can tell the state of the art small open models is qwen3.6 and gemma 4, yet they rarely appear in benchmarks - even ones made recently

But even if distribution is required to trigger the law - you are free to distribute fine-tunes of most open weight models.

Use pi.dev as your agent harness, and point it at ollama or whatever you’re running your model in.

Since qwen3.6-35b-a3b is a “MoE” model (indicated by the a3b part), you can get away with a lot less VRAM and a slower GPU. For me, 35b-a3b performs about as well as a 12b “dense” model. You’ll still need enough RAM+VRAM to hold the entire model - otherwise it’ll be super slow as it loads pieces of the model off disk for every token. Loading from RAM to VRAM is slower than having the whole thing in VRAM but better than involving your ssd.