GPU comparison
Mac Studio M2 Ultra (192GB) vs NVIDIA RTX 3090 (24GB) for local LLMs
How the two stack up for running open-source LLMs locally — memory, bandwidth, price, and how many of the 22 tracked models each one can run.
| Mac Studio M2 Ultra (192GB) | NVIDIA RTX 3090 (24GB) | |
|---|---|---|
| Memory | 192.0 GB | 24.0 GB |
| Bandwidth | 800 GB/s | 936 GB/s |
| Price (approx) | $5,599 | $899 |
| LLMs it runs | 20 of 22 | 16 of 22 |
| Best model it runs | Hunyuan Hy3 (MoE) · 48–71 tok/s | Gemma 4 31B · 31–47 tok/s |
The Mac Studio M2 Ultra (192GB) runs 4 more of the tracked models (20 vs 16), thanks to its 192.0 GB of memory. The NVIDIA RTX 3090 (24GB) has more memory bandwidth (936 GB/s), so it generates tokens faster at the same model and quant.