Written by Jakub Rusinowski · Last updated August 15, 2026
Yes, but it is tight
Yes, but it is tight — Granite 4.0 Small-H 32B-A9B at Q6_K needs about 28.9 GB of the 32 GB usable on Beelink SER9 (Ryzen AI 9, 32 GB), leaving only ~3.1 GB before the runtime starts swapping. Expect ~22 tok/s (estimated), with room for about 16,384 tokens of context.
Confidence: low · Recommended quantization: Q6_K · Estimated speed: ~22 tok/s
| 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) |
| Quant | Memory needed | Fits 32 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 66.6 GB | ✗ No | — | — | 64 GB |
| Q8_0 | 36.6 GB | ✗ No | — | — | 34 GB |
| Q6_K | 28.9 GB | ✓ Yes | 16K | ~22 tok/s | 26.2 GB |
| Q5_K_M | 25.3 GB | ✓ Yes | 32K | ~24.8 tok/s | 22.7 GB |
| Q4_K_M | 22 GB | ✓ Yes | 32K | ~28.3 tok/s | 19.3 GB |
| Q3_K_M | 16.3 GB | ✓ Yes | 64K | ~37 tok/s | 13.6 GB |
| Q2_K | 13.2 GB | ✓ Yes | 64K | ~44.5 tok/s | 10.5 GB |
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Yes, but it is tight — Granite 4.0 Small-H 32B-A9B at Q6_K needs about 28.9 GB of the 32 GB usable on Beelink SER9 (Ryzen AI 9, 32 GB), leaving only ~3.1 GB before the runtime starts swapping. Expect ~22 tok/s (estimated), with room for about 16,384 tokens of context.
Q6_K — it needs about 28.9 GB of the 32 GB available, downloads as roughly 26.2 GB, and runs at an estimated 22 tokens/sec with up to 16K of context.
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
← Can I Run It? | IBM Granite 4.0 model page | Check your hardware