Less. Probably 2-3K if you build right. Qwen 3.8 27B on constrained tasks is Opus 4.6-ish to me, it just doesn’t know enough, but when task is laid out just gets it done.
Comes down to how much of the ambiguity we expect out of the model.
Comes down to how much of the ambiguity we expect out of the model.
Need 64GB vram, 900+ GB/s speed, and a lot of system ram (128GB). Seems feasible. Hmm.
Maybe older GPUs work.