community-verified · measured pool data
Can the M5 Max 128GB run DeepSeek-V4-Flash?
Yes — community-measured at 19 tok/s (5 runs at 2-bit).
Verified single-stream runs on the M5 Max 128GB from the localmaxxing public pool. Every number on this page is measured — nothing here is estimated.
| Quantization | Median tok/s | TTFT (ms) | Peak VRAM | Max context tested | Engines |
|---|---|---|---|---|---|
| 2-bit | 19 | 70150 | — | 128,000 | llama.cpp |