Yes, just — 2.4 GB spare

Olmo 3.1 32B Instruct at Q4_K_M fits your GeForce RTX 5090 Laptop entirely on the GPU at 8K context, at an estimated 30 tokens per second. Past 47K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.

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.

What hardware do I need for Olmo 3.1 32B Instruct? →

The VRAM budget

weights 18.1 GB

Weights 18.1 GB KV cache @ 8K 1.25 GB Runtime overhead 0.6 GB Free 2.4 GB of 22.4 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
Q8_0	31.9 GB	33.7 GB	—	~2.9	−0.1% ppl	11.3 GB over
Q6_K	24.6 GB	26.4 GB	—	~6.7	−0.4% ppl	4.0 GB over
Q5_K_M	21.3 GB	23.1 GB	2K	~17	−0.8% ppl	0.7 GB over
Q4_K_M	18.1 GB	20.0 GB	47K	30	−1.9% ppl	Recommended
Q3_K_M	14.7 GB	16.5 GB	64K	37	−5.4% ppl	Long context
Q2_K	12.6 GB	14.4 GB	64K	43	−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

Ollama llama.cpp LM Studio vLLM

terminal

$ 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.

02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.

03Only 2.4 GB is spare, so a long system prompt can still push it over. Its real ceiling here is 47K context.

See all models for this rig Compare with another model