2benchesM2 Max
A15 · M2 · Avalanche / Blizzard · 5 nm · First available 2023-01-24
Every configuration
| Configuration | CPU core layout | Bandwidth | GB7 CPU | GB7 Metal | Prefill Q4_0 tok/s | Decode Q4_0 tok/s |
|---|---|---|---|---|---|---|
| 12 CPU / 30 GPU (binned) | 8P + 4E | 400 GB/s | 17,071 | — | 537.6 | 61 |
| 12 CPU / 38 GPU (full chip) | 8P + 4E | 400 GB/s | 17,234 | 142,917 | 671.3 | 66 |
All benchmarks
| Benchmark | Median |
|---|---|
| Geekbench 7 · Multi-core | 17,071 (17,071–17,397, 3 sources) |
| Geekbench 7 · Metal GPU | 142,917 (140,491–145,343, 2 sources) |
| Geekbench AI · GPU · Single | — |
| Geekbench AI · GPU · Half | — |
| Geekbench AI · GPU · Quantized | — |
| Geekbench AI · Neural Engine · Single | — |
| Geekbench AI · Neural Engine · Half | — |
| Geekbench AI · Neural Engine · Quantized | — |
| Prefill 512 · F16 | 678.1 (600.5–755.7, 2 sources) |
| Decode 128 · F16 | 24.4 (24.2–24.7, 2 sources) |
| Prefill 512 · Q8_0 | 609 (540.2–677.9, 2 sources) |
| Decode 128 · Q8_0 | 40.9 (40–41.8, 2 sources) |
| Prefill 512 · Q4_0 | 604.5 (537.6–671.3, 2 sources) |
| Decode 128 · Q4_0 | 63.5 (61–66, 2 sources) |
Similar performance
Nearest measured configurations by Geekbench 7 CPU and Metal. Ratios above 1 are faster.
12 CPU / 38 GPU (full chip)
- M3 Max 14 CPU / 30 GPU (binned), 2023-11: CPU 1.28×, GPU 0.92×
- M3 Max 16 CPU / 40 GPU (full chip), 2023-11: CPU 1.42×, GPU 1.14×
- M4 Pro 12 CPU / 16 GPU (binned), 2024-11: CPU 1.26×, GPU 0.72×
- M4 Pro 14 CPU / 20 GPU (full chip), 2024-11: CPU 1.44×, GPU 0.81×
- M4 Max 14 CPU / 32 GPU (binned), 2024-11: CPU 1.48×, GPU 1.18×
- M2 Pro 12 CPU / 19 GPU (full chip), 2023-01: CPU 0.96×, GPU 0.58×