Can I Run Llama 3.3 on Beelink SER9 (Ryzen AI 9, 32 GB)?
Written by Jakub Rusinowski · Last updated December 8, 2024
Technically yes, but not recommended
It loads, but it is not worth running — Llama 3.3 70B Instruct at Q2_K fits in Beelink SER9 (Ryzen AI 9, 32 GB)'s 32 GB, yet the memory bandwidth limits it to ~3.7 tok/s (estimated), well below usable interactive speed.
Confidence: high · Recommended quantization: Q2_K · Estimated speed: ~3.7 tok/s
Get a personalized upgrade path →
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
Beelink SER9 (Ryzen AI 9, 32 GB) — what it gives a model
| Usable memory for models | 32 GB |
| Memory bandwidth | 120 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) |
Llama 3.3 on Beelink SER9 (Ryzen AI 9, 32 GB): memory by quantization
| Quant | Memory needed | Fits 32 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 143.5 GB | ✗ No | — | — | 140 GB |
| Q8_0 | 77.9 GB | ✗ No | — | — | 74.4 GB |
| Q6_K | 60.9 GB | ✗ No | — | — | 57.4 GB |
| Q5_K_M | 53.1 GB | ✗ No | — | — | 49.6 GB |
| Q4_K_M | 45.7 GB | ✗ No | — | — | 42.3 GB |
| Q3_K_M | 33.3 GB | ✗ No | — | — | 29.8 GB |
| Q2_K | 26.5 GB | ✓ Yes | 16K | ~3.7 tok/s | 23 GB |
What to watch out for
- At ~3.7 tok/s this loads but is too slow for interactive use — expect roughly 16 seconds per 60 tokens.
- Q2_K is the only quantization that fits, and it is a heavily degraded one — expect noticeably worse output than the same model at Q4_K_M. A smaller model at Q4 is usually the better trade.
- 1 larger variant of Llama 3.3 does not fit and would need CPU offload or different hardware.
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 120 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 computed from this model's published attention configuration.
FAQ
Can I run Llama 3.3 on Beelink SER9 (Ryzen AI 9, 32 GB)?
It loads, but it is not worth running — Llama 3.3 70B Instruct at Q2_K fits in Beelink SER9 (Ryzen AI 9, 32 GB)'s 32 GB, yet the memory bandwidth limits it to ~3.7 tok/s (estimated), well below usable interactive speed.
Which quantization of Llama 3.3 should I use on Beelink SER9 (Ryzen AI 9, 32 GB)?
Q2_K — it needs about 26.5 GB of the 32 GB available, downloads as roughly 23 GB, and runs at an estimated 3.7 tokens/sec with up to 16K of context.
What limits Llama 3.3 on Beelink SER9 (Ryzen AI 9, 32 GB)?
Memory bandwidth. The model fits, but at 120 GB/s it can only be read fast enough for roughly 3.7 tokens/sec.
Which runtime should I use?
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Other Models on Beelink SER9 (Ryzen AI 9, 32 GB)
- Magistral Small on Beelink SER9 (Ryzen AI 9, 32 GB)
- MiniCPM-V on Beelink SER9 (Ryzen AI 9, 32 GB)
- Ministral on Beelink SER9 (Ryzen AI 9, 32 GB)
- Ministral 3 on Beelink SER9 (Ryzen AI 9, 32 GB)
- Mistral Family on Beelink SER9 (Ryzen AI 9, 32 GB)
Llama 3.3 on GPUs
- Llama 3.3 on NVIDIA GeForce RTX 5090
- Llama 3.3 on NVIDIA GeForce RTX 5080
- Llama 3.3 on NVIDIA GeForce RTX 5070 Ti
- Llama 3.3 on NVIDIA GeForce RTX 5070
What This Model Is Good At
Model & Tools
← Can I Run It? | Llama 3.3 model page | Check your hardware