I got fed up with Claude code limits and have been using a combination of qwen3-coder, gemma4, and qwen3-vl locally. Gets me 90% of the way there and CC is still around for now if I need it.

Btw even at insane markups $200/mo means GPUs break even pretty fast.

Which harness and how which GPU?

the hardware ROI is insane right now tbh. a $200/mo sub is literally paying off a second gpu in less than a year.