2benchesM3 Max
A16 / A17 · M3 · Everest / Sawtooth family · 3 nm · First available 2023-11-07
Every configuration
| Configuration | CPU core layout | Bandwidth | GB7 CPU | GB7 Metal | Prefill Q4_0 tok/s | Decode Q4_0 tok/s |
|---|---|---|---|---|---|---|
| 14 CPU / 30 GPU (binned) | 10P + 4E | 300 GB/s | 22,035 | 131,592 | 567.6 | 56.6 |
| 16 CPU / 40 GPU (full chip) | 12P + 4E | 400 GB/s | 24,422 | 163,381 | 759.7 | 66.3 |
All benchmarks
| Benchmark | Median |
|---|---|
| Geekbench 7 · Multi-core | 23,111 (22,017–24,675, 4 sources) |
| Geekbench 7 · Metal GPU | 146,062 (130,616–167,206, 4 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 | 684.3 (589.4–779.2, 2 sources) |
| Decode 128 · F16 | 22.3 (19.5–25.1, 2 sources) |
| Prefill 512 · Q8_0 | 662 (566.4–757.6, 2 sources) |
| Decode 128 · Q8_0 | 38.5 (34.3–42.8, 2 sources) |
| Prefill 512 · Q4_0 | 663.7 (567.6–759.7, 2 sources) |
| Decode 128 · Q4_0 | 61.5 (56.6–66.3, 2 sources) |
Similar performance
Nearest measured configurations by Geekbench 7 CPU and Metal. Ratios above 1 are faster.
14 CPU / 30 GPU (binned)
- M4 Pro 14 CPU / 20 GPU (full chip), 2024-11: CPU 1.12×, GPU 0.88×
- M4 Pro 12 CPU / 16 GPU (binned), 2024-11: CPU 0.98×, GPU 0.78×
- M2 Max 12 CPU / 38 GPU (full chip), 2023-01: CPU 0.78×, GPU 1.09×
- M4 Max 14 CPU / 32 GPU (binned), 2024-11: CPU 1.15×, GPU 1.28×
- M5 Pro 15 CPU / 16 GPU (binned), 2026-03: CPU 1.34×, GPU 0.95×
- M5 Pro 18 CPU / 20 GPU (full chip), 2026-03: CPU 1.52×, GPU 1.03×
16 CPU / 40 GPU (full chip)
- M4 Max 14 CPU / 32 GPU (binned), 2024-11: CPU 1.04×, GPU 1.03×
- M4 Max 16 CPU / 40 GPU (full chip), 2024-11: CPU 1.18×, GPU 1.25×
- M5 Pro 15 CPU / 16 GPU (binned), 2026-03: CPU 1.21×, GPU 0.76×
- M2 Ultra 24 CPU / 76 GPU (full chip), 2023-06: CPU 1.16×, GPU 1.36×
- M4 Pro 14 CPU / 20 GPU (full chip), 2024-11: CPU 1.01×, GPU 0.7×
- M5 Pro 18 CPU / 20 GPU (full chip), 2026-03: CPU 1.37×, GPU 0.83×