Phi-4 on NVIDIA GeForce RTX 4090 · 24GB VRAM

✅ Runs comfortably — Phi-4 14B @ Q8_0
Phi-4 14B at Q8_0 needs ~16.8 GB (weights 14.5 GB + KV 1.6 GB + overhead 743 MB @ 8K ctx) of your 21.6 GB usable VRAM — plenty of headroom. Expect ~30–45 tok/s (fast).
context @ Q8_0: comfortable to 8K · runs to its full 16K
quantneedsspeed
✓ Q8_016.8 GB~30–45 tok/s (fast)
✓ Q5_K_M11.9 GB~40–70 tok/s (very fast)
✓ Q4_K_M10.5 GB~50–80 tok/s (very fast)
✓ Q3_K_M8.9 GB~60–100 tok/s (very fast)
$runlocal install phi4:14b
⚠ NVIDIA support is best-effort in v0.1 — verify before relying on it.
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
Phi-4: Runs comfortably
for your own README — links back here

Same model, other machines: Apple M1 · Apple M1 Pro · Apple M2 · Apple M3 Pro · Apple M2 Pro · Apple M4 Pro

Also runs on a NVIDIA GeForce RTX 4090: DeepSeek-R1 (distill) · Qwen3 · Mistral Small 3 · Qwen2.5 · Gemma 3

Best models for a NVIDIA GeForce RTX 4090 · 24GB VRAM →

Check any combo yourself: open the checker →