QwQ 32B on NVIDIA GeForce RTX 5080 · 16GB VRAM

⚠️ Partial GPU offload — QwQ 32B @ Q4_K_M
QwQ 32B 32B at Q4_K_M needs ~21.4 GB (weights 18.5 GB + KV 2.0 GB + overhead 947 MB @ 8K ctx) of your 14.4 GB usable VRAM — some layers spill to system RAM. Expect ~6–9 tok/s (slow).
context @ Q4_K_M: runs to 16K · won't fit past that on this rig
quantneedsspeed
✗ Q8_036.1 GB—
~ Q5_K_M24.7 GB—
~ Q4_K_M21.4 GB~6–9 tok/s (slow)
$runlocal install qwq:32b
⚠ 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
QwQ 32B: Partial GPU offload
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 5080: DeepSeek-R1 (distill) · Qwen2.5 · Qwen3 · Gemma 4 · Gemma 3

Best models for a NVIDIA GeForce RTX 5080 · 16GB VRAM →

Architecturally similar

These share QwQ 32B's KV-cache geometry — they size memory the same way, so here's how they fit on a NVIDIA GeForce RTX 5080 · 16GB VRAM too:

modelsharesverdict here
Llama 3.1 8kv/128hd ✅ Runs comfortably
Qwen2.5-Coder 8kv/128hd ✅ Runs (tight)
Llama 3.2 8kv/128hd ✅ Runs comfortably
Llama 3.3 8kv/128hd ❌ Won't fit
Mistral 7B 8kv/128hd ✅ Runs comfortably
Mistral Small 3 8kv/128hd ✅ Runs (tight)
Mistral Nemo 8kv/128hd ✅ Runs (tight)
Codestral 8kv/128hd ✅ Runs (tight)

…and 8 more — see the QwQ 32B page.

Check any combo yourself: open the checker →