gpt-oss on NVIDIA GeForce RTX 5090 · 32GB VRAM

✅ Runs comfortably — gpt-oss 20B-A3.6B (MoE) @ MXFP4
gpt-oss 20B-A3.6B (MoE) at MXFP4 needs ~12.2 GB (weights 11.3 GB + KV 384 MB + overhead 577 MB @ 8K ctx) of your 28.8 GB usable VRAM — plenty of headroom. Expect ~170–280 tok/s (very fast).
context @ MXFP4: comfortable to 128K
quantneedsspeed
✓ MXFP412.2 GB~170–280 tok/s (very fast)
$runlocal install gpt-oss:20b
⚠ 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
gpt-oss: 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 gpt-oss'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
Llama 3.2 8kv/64hd ✅ Runs comfortably

Check any combo yourself: open the checker →