GPU comparison
MacBook Air/Pro M5 (24GB) vs NVIDIA RTX 3060 (12GB) 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 (24GB) | NVIDIA RTX 3060 (12GB) | |
|---|---|---|
| Memory | 24.0 GB | 12.0 GB |
| Bandwidth | 153 GB/s | 360 GB/s |
| Price (approx) | $1,299 | $329 |
| LLMs it runs | 45 of 64 | 27 of 64 |
| Best model it runs | Qwen3.8 27B · 9–13 tok/s | Devstral Small 2 24B · 26–39 tok/s |
The MacBook Air/Pro M5 (24GB) runs 18 more of the tracked models (45 vs 27), thanks to its 24.0 GB of memory. The NVIDIA RTX 3060 (12GB) has more memory bandwidth (360 GB/s), so it generates tokens faster at the same model and quant.