local intelligence

can it run

Can the NVIDIA RTX 2070 run gemma3 12B?

OFFLOAD PARTIAL

9.6 GB vs 8 GB usable — partial CPU offload, expect large speed loss

8GB · Turing TU106 · model file 8.1GB · 12B

~27–40 tok/s

computed roofline: 448 GB/s × 0.48–0.72 efficiency window / 8.1GB (Q4_K_M) — bands, never points.

Run gemma3 12B on instead

Tesla P100 (16GB, used) — cheapest catalog machine that runs gemma3 12B fully ($135)

RTX 3090 (24GB, used) — upgrade path machine

every model the NVIDIA RTX 2070 can run — full list

explore the full catalog84 machines indexed · live prices · what each one can run
2026-09-17 · ← all models on the NVIDIA RTX 2070 · fit = weights + 4k KV vs usable memory