▲ local intelligence

can it run

Can the NVIDIA RTX A5000 run phi4 14B?

FULL

9.9 GB needed of 24 GB usable — headroom for context

24GB · Ampere · model file 9.1GB · 14B

~39–64 tok/s

computed roofline: 768 GB/s × 0.46–0.76 efficiency window / 9.1GB (Q4_K_M) — bands, never points.

Run phi4 14B on instead

→ Tesla M60 (16GB, used) — cheapest catalog machine that runs phi4 14B fully ($35)

→ rtx-a6000 — the next step up from NVIDIA RTX A5000

→ every model the NVIDIA RTX A5000 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 RTX A5000 · fit = weights + 4k KV vs usable memory