Side by side
| M5 | M5 Pro | |
|---|---|---|
| Memory sizes | 16, 24, 32 GB | 24, 48, 64 GB |
| Memory bandwidth | 153 GB/s | 307 GB/s |
| For models, at most | about 24 GB | about 48 GB |
| Models that run well | 28 of 73 | 34 of 73 |
| MLX seat class | Core | Pro Latest |
| Found in | MacBook Pro (14-inch, M5), MacBook Air (13-inch, M5), MacBook Air (15-inch, M5) | MacBook Pro (14-inch, M5 Pro or M5 Max), MacBook Pro (16-inch, M5 Pro or M5 Max), Mac mini (M6 or M5 Pro) |
Speed with popular models
Tokens a second for one conversation, each chip at its largest memory size (32 GB and 64 GB), with the best version of the model that fits.
| Model | M5 | M5 Pro |
|---|---|---|
| Llama 3.1 8B Instruct8B | 11–19 6-bit | 16–29 8-bit |
| gpt-oss-20b21B · 3.6B active | 36–64 native 4.25-bit | 73–129 native 4.25-bit |
| Qwen3.6 35B A3B35B · 3B active | 41–72 4-bit | 43–77 8-bit |
| Qwen3.8 27B27B | 3–6 6-bit | 5–9 8-bit |
| Llama 3.3 70B Instruct70B | 3–6 2-bit lower quality | 4–6 4-bit |
| gpt-oss-120b117B · 5.1B active | too big | 86–154 2-bit lower quality |
| Qwen3.5 122B A10B122B · 10B active | too big | 45–79 2-bit lower quality |
| DeepSeek V4 Flash 0731284B · 13B active | too big | too big |
Questions
- Is the M5 Pro faster than the M5 for LLMs?
- The M5 Pro reads memory about 101% faster than the M5 (307 vs 153 GB/s), so it writes about 101% faster with any model that fits both. Reading a long prompt depends on GPU compute instead, which this comparison doesn’t cover.
- Which can run bigger models, the M5 or the M5 Pro?
- The M5 Pro: The M5 Pro goes up to 64 GB of memory, the M5 to 32 GB. At its largest it runs 34 of 73 open models well, against 28.
More comparisons
The M5 for local AI · The M5 Pro for local AI · Compare every model · The best Mac for local LLMs
How it’s worked out
Memory: the model’s parameters at 8, 6, 4, 3 or 2 bits per weight plus MLX’s group scales (half a bit per weight), or its native format, with room for the context on top. By default macOS lets the GPU use about 75% of unified memory. Speed: memory bandwidth divided by the bytes read per token is the ceiling for one conversation; in llama.cpp’s public Apple Silicon benchmarks real runs reach about 45% to 80% of it, which is the range shown. These are estimates, not measurements: measure your own Mac on MLX Bench. Model figures from each official model card, chip figures from Apple’s Tech Specs. Checked 2026-10-19. Same data as mlx.app/compare.