Can I Run Poolside Laguna XS 2.1 on Beelink SER9 (Ryzen AI 9, 32 GB)?

Written by Jakub Rusinowski · Last updated August 15, 2026

Yes, but it is tight

Yes, but it is tight — Laguna XS 2.1 33B-A3B at Q6_K needs about 29.7 GB of the 32 GB usable on Beelink SER9 (Ryzen AI 9, 32 GB), leaving only ~2.3 GB before the runtime starts swapping. Expect ~50.3 tok/s (estimated), with room for about 16,384 tokens of context.

Confidence: low · Recommended quantization: Q6_K · Estimated speed: ~50.3 tok/s

Beelink SER9 (Ryzen AI 9, 32 GB) — what it gives a model

Usable memory for models	32 GB
Memory bandwidth	256 GB/s
Form factor	Mini PC
Operating system	Windows or Linux
Memory upgradeable	Yes
Price	$859 (lib/data/ai-stations.ts (street price), checked 2026-07-06)

Poolside Laguna XS 2.1 on Beelink SER9 (Ryzen AI 9, 32 GB): memory by quantization

Quant	Memory needed	Fits 32 GB?	Max context	Est. speed	Download
F16	68.6 GB	✗ No	—	—	66 GB
Q8_0	37.7 GB	✗ No	—	—	35.1 GB
Q6_K	29.7 GB	✓ Yes	16K	~50.3 tok/s	27.1 GB
Q5_K_M	26 GB	✓ Yes	32K	~55.2 tok/s	23.4 GB
Q4_K_M	22.6 GB	✓ Yes	32K	~60.7 tok/s	19.9 GB
Q3_K_M	16.7 GB	✓ Yes	64K	~72.9 tok/s	14.1 GB
Q2_K	13.5 GB	✓ Yes	64K	~82 tok/s	10.8 GB

What to watch out for

This model does not publish its full attention configuration, so the KV-cache share of these figures is inferred from its parameter count rather than computed exactly.

Beelink SER9 32 GB limitations

Shares system memory with the iGPU, so the usable model budget is well below the nominal 32 GB.
Memory bandwidth, not capacity, is the limit here — expect single-digit tokens/sec on large models.

Recommended setup

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

How these numbers are calculated

Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
32 GB unified memory at 256 GB/s, shared between CPU and GPU.
Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
KV cache is inferred from the parameter count because this model does not publish its full attention configuration.

FAQ