GPU comparison
MacBook Air/Pro M5 (16GB) vs MacBook Pro M5 Max (36GB) for local LLMs
How the two stack up for running open-source LLMs locally — memory, bandwidth, price, and how many of the 64 tracked models each one can run.
| MacBook Air/Pro M5 (16GB) | MacBook Pro M5 Max (36GB) | |
|---|---|---|
| Memory | 16.0 GB | 36.0 GB |
| Bandwidth | 153 GB/s | 460 GB/s |
| Price (approx) | $1,099 | $3,599 |
| LLMs it runs | 27 of 64 | 47 of 64 |
| Best model it runs | Devstral Small 2 24B · 10–15 tok/s | Qwen3.8 27B · 13–20 tok/s |
The MacBook Pro M5 Max (36GB) runs 20 more of the tracked models (47 vs 27), thanks to its 36.0 GB of memory. The MacBook Pro M5 Max (36GB) has more memory bandwidth (460 GB/s), so it generates tokens faster at the same model and quant.