▲ local intelligence

can it run

Can the NVIDIA GeForce RTX 3050 8GB run gemma3 12B?

OFFLOAD PARTIAL

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

8GB · Ampere · model file 8.1GB · 12B

~15–24 tok/s

computed roofline: 224 GB/s × 0.55–0.88 efficiency window / 8.1GB (Q4_K_M) — bands, never points.

Run gemma3 12B on instead

→ Tesla M60 (16GB, used) — cheapest catalog machine that runs gemma3 12B fully ($35)

→ rtx-3060-ti — the next step up from NVIDIA GeForce RTX 3050 8GB

→ every model the NVIDIA GeForce RTX 3050 8GB can run — full list

→explore the full catalog91 machines indexed · live prices · what each one can run
2026-10-01 · ← all models on the NVIDIA GeForce RTX 3050 8GB · fit = weights + 4k KV vs usable memory