Gemma 4 on NVIDIA GeForce RTX 5090 · 32GB VRAM

✅ Runs comfortably — Gemma 4 12B @ Q8_0
Gemma 4 12B at Q8_0 needs ~15.8 GB (weights 12.2 GB + KV 3.0 GB + overhead 626 MB @ 8K ctx) of your 28.8 GB usable VRAM — plenty of headroom. Expect ~50–80 tok/s (very fast).
context @ Q8_0: comfortable to 16K · runs to 64K · won't fit past that on this rig
quantneedsspeed
✓ Q8_015.8 GB~50–80 tok/s (very fast)
✓ Q6_K13.0 GB~60–95 tok/s (very fast)
✓ Q5_K_M11.7 GB~65–110 tok/s (very fast)
✓ Q4_K_M10.6 GB~70–120 tok/s (very fast)
⚠ 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
Gemma 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 5090: DeepSeek-R1 (distill) · Qwen2.5 · Qwen3 · Llama 3.1 · Gemma 3

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

Architecturally similar

These share Gemma 4's KV-cache geometry — they size memory the same way, so here's how they fit on a NVIDIA GeForce RTX 5090 · 32GB VRAM too:

modelsharesverdict here
Gemma 2 8kv/256hd ✅ Runs comfortably

Check any combo yourself: open the checker →