2benchesM4 Pro
A18 · M4 · 2024 CPU family · 3 nm · First available 2024-11-08
Every configuration
| Configuration | CPU core layout | Bandwidth | GB7 CPU | GB7 Metal | Prefill Q4_0 tok/s | Decode Q4_0 tok/s |
|---|---|---|---|---|---|---|
| 12 CPU / 16 GPU (binned) | 8P + 4E | 273 GB/s | 21,691 | 102,220 | 364.1 | 49.6 |
| 14 CPU / 20 GPU (full chip) | 10P + 4E | 273 GB/s | 24,755 | 115,152 | 439.8 | 50.7 |
All benchmarks
| Benchmark | Median |
|---|---|
| Geekbench 7 · Multi-core | 24,714 (21,506–24,910, 5 sources) |
| Geekbench 7 · Metal GPU | 113,068 (102,049–115,481, 5 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 | 422.8 (381.1–464.5, 2 sources) |
| Decode 128 · F16 | 17.2 (17.2–17.2, 2 sources) |
| Prefill 512 · Q8_0 | 408.4 (367.1–449.6, 2 sources) |
| Decode 128 · Q8_0 | 30.6 (30.5–30.7, 2 sources) |
| Prefill 512 · Q4_0 | 401.9 (364.1–439.8, 2 sources) |
| Decode 128 · Q4_0 | 50.2 (49.6–50.7, 2 sources) |
Similar performance
Nearest measured configurations by Geekbench 7 CPU and Metal. Ratios above 1 are faster.
12 CPU / 16 GPU (binned)
- M3 Max 14 CPU / 30 GPU (binned), 2023-11: CPU 1.02×, GPU 1.29×
- M2 Pro 12 CPU / 19 GPU (full chip), 2023-01: CPU 0.76×, GPU 0.81×
- M5 Pro 15 CPU / 16 GPU (binned), 2026-03: CPU 1.36×, GPU 1.22×
- M3 Pro 12 CPU / 18 GPU (full chip), 2023-11: CPU 0.78×, GPU 0.75×
- M2 Max 12 CPU / 38 GPU (full chip), 2023-01: CPU 0.79×, GPU 1.4×
- M5 10 CPU / 10 GPU (full chip), 2025-10: CPU 0.79×, GPU 0.69×
14 CPU / 20 GPU (full chip)
- M3 Max 14 CPU / 30 GPU (binned), 2023-11: CPU 0.89×, GPU 1.14×
- M5 Pro 15 CPU / 16 GPU (binned), 2026-03: CPU 1.2×, GPU 1.08×
- M5 Pro 18 CPU / 20 GPU (full chip), 2026-03: CPU 1.35×, GPU 1.18×
- M3 Max 16 CPU / 40 GPU (full chip), 2023-11: CPU 0.99×, GPU 1.42×
- M4 Max 14 CPU / 32 GPU (binned), 2024-11: CPU 1.03×, GPU 1.46×
- M2 Max 12 CPU / 38 GPU (full chip), 2023-01: CPU 0.7×, GPU 1.24×