Phi-3.5-mini on NVIDIA GeForce RTX 5090 · 32GB VRAM

✅ Runs comfortably — Phi-3.5-mini 3.8B @ Q8_0
Phi-3.5-mini 3.8B at Q8_0 needs ~7.3 GB (weights 3.8 GB + KV 3.0 GB + overhead 512 MB @ 8K ctx) of your 28.8 GB usable VRAM — plenty of headroom. Expect ~110–180 tok/s (very fast).
context @ Q8_0: comfortable to 32K · runs to 64K · won't fit past that on this rig
quantneedsspeed
✓ Q8_07.3 GB~110–180 tok/s (very fast)
✓ Q5_K_M6.0 GB~130–220 tok/s (very fast)
✓ Q4_K_M5.7 GB~140–230 tok/s (very fast)
$runlocal install phi3.5:3.8b
⚠ 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-3.5-mini: 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 5090: DeepSeek-R1 (distill) · Qwen2.5 · Qwen3 · Llama 3.1 · Gemma 3

Best models for a NVIDIA GeForce RTX 5090 · 32GB VRAM →

Check any combo yourself: open the checker →