Only with CPU offload

Qwen3 32B at Q4_K_M needs 21.0 GB but only 14.4 GB is addressable, so about 36% of the layers would stream from system RAM at roughly 60 GB/s. Expect around 4.7 tokens per second — usable for batch work, painful for chat.

64% on GPU 8K context Q4_K_M · 18.4 GB Apache 2.0 Released Apr 2025

The classic 24 GB target, and still the strongest local translator under 70B. Qwen3.8 27B is smaller and better at everything else.

What hardware do I need for Qwen3 32B? →

Fits instead: Qwen3 14B (10.2 GB) · Qwen3 8B (6.3 GB)

The VRAM budget

weights 18.4 GB

Weights 18.4 GB KV cache @ 8K 2.00 GB Runtime overhead 0.6 GB Over budget 6.6 GB past 14.4 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
Q8_0	32.5 GB	35.1 GB	—	~1.7	−0.1% ppl	20.7 GB over
Q6_K	25.0 GB	27.6 GB	—	~2.5	−0.4% ppl	13.2 GB over
Q5_K_M	21.7 GB	24.3 GB	—	~3.3	−0.8% ppl	9.9 GB over
Q4_K_M	18.4 GB	21.0 GB	—	~4.7	−1.9% ppl	6.6 GB over
Q3_K_M	14.9 GB	17.5 GB	—	~8.6	−5.4% ppl	3.1 GB over
Q2_K	12.8 GB	15.4 GB	4K	~17	−15% ppl	1.0 GB over

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.

How to run it

Ollama llama.cpp LM Studio vLLM

terminal

$ llama-server \
    -hf Qwen/Qwen3-32B:Q4_K_M \
    -c 8192 -ngl 40

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 18.4 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.

03If it falls back to CPU silently, drop the context first, then step down a quantisation.

See all models for this rig Compare with another model