Can Mac M3 Pro 36G run Arete 27B?
Comfortable: GGUF Q5_K_M needs 21.9 GB at 4K, batch 1; headroom 5.1 GB.
Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.
Measured runs
No measured run for this pair. Sizing is an estimate, not a runtime test.
Quant memory at 4K
| Quant | Total GB | Verdict | Weight sizing |
|---|---|---|---|
| GGUF Q3_K_M | 14.0 | Comfortable | Verified file |
| GGUF Q4_K_M | 18.3 | Comfortable | Verified file |
| GGUF Q5_K_M | 21.9 | Comfortable | Verified file |
| GGUF Q8_0 | 30.2 | Does not fit | Verified file |
Context capacity · GGUF Q5_K_M
| Context | Total GB | Verdict |
|---|---|---|
| 4K | 21.9 | Comfortable |
| 16K | 22.7 | Comfortable |
| 32K | 23.8 | Tight |
| 128K | 30.4 | Does not fit |
Generation speed
GGUF Q5_K_M: 6–7 tok/s · Derived range