Yes — with 5.6 GB to spare
Olmo 3.1 32B Instruct at Q4_K_M fits your CPU only · DDR4 dual-channel entirely on the GPU at 8K context, at an estimated 1.2 tokens per second. There is room for its full 64K window.
Fully on GPU
8K context
Q4_K_M · 18.1 GB
Apache 2.0
Released 10 Dec 2025
The largest fully open model you can audit end to end. Q4 fits 24 GB, tightly.
The VRAM budget
weights 18.1 GB
Weights 18.1 GB
KV cache @ 8K 1.25 GB
Runtime overhead 0.6 GB
Free 5.6 GB of 25.6 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 31.9 GB | 33.7 GB | — | ~0.7 | −0.1% ppl | 8.1 GB over |
| Q6_K | 24.6 GB | 26.4 GB | 1K | ~0.9 | −0.4% ppl | 0.8 GB over |
| Q5_K_M | 21.3 GB | 23.1 GB | 47K | 1.0 | −0.8% ppl | Long context |
| Q4_K_M | 18.1 GB | 20.0 GB | 64K | 1.2 | −1.9% ppl | Recommended |
| Q3_K_M | 14.7 GB | 16.5 GB | 64K | 1.5 | −5.4% ppl | Long context |
| Q2_K | 12.6 GB | 14.4 GB | 64K | 1.7 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model interleaves sliding-window layers (4096 tokens, 16 of 64 layers global), which is why its cache barely grows with context.
How to run it
$ llama-server \
-hf allenai/Olmo-3.1-32B-Instruct:Q4_K_M \
-c 8192 -ngl 99
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.
01Download is 18.1 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02With no GPU, thread count matters more than clock. Start at one thread per physical core.
03There is room to go to the model's full 64K context on this card.