can it run
6.7 of 8 GB usable — keep context modest
8GB · Turing TU106 · model file 4.7GB · 7B
computed roofline: 448 GB/s × 0.48–0.72 efficiency window / 4.7GB (Q4_K_M) — bands, never points.
→ Tesla M60 (16GB, used) — cheapest catalog machine that runs llava 7B fully ($35)
→ rtx-3060-12 — the next step up from NVIDIA GeForce RTX 2060 Super
→ every model the NVIDIA GeForce RTX 2060 Super can run — full list
→explore the full catalog91 machines indexed · live prices · what each one can run