Can Mac M5 Ultra 256G run K2 Horizon 375B-A23B?
Comfortable: GGUF Q2_K needs 154.1 GB at 4K, batch 1; headroom 37.9 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 Q2_K | 154.1 | Comfortable | Verified file |
| GGUF Q3_K_M | 191.7 | Tight | Verified file |
| GGUF Q4_K_M | 249.7 | Does not fit | Verified file |
| GGUF Q8_0 | 422.0 | Does not fit | Verified file |
Context capacity · GGUF Q2_K
| Context | Total GB | Verdict |
|---|---|---|
| 4K | 154.1 | Comfortable |
| 16K | 157.2 | Comfortable |
| 32K | 161.4 | Comfortable |
| 128K | 186.6 | Tight |
Generation speed
GGUF Q2_K: 142 tok/s · Bandwidth ceiling (not measured)