GPU comparison
NVIDIA RTX 3090 (24GB) vs NVIDIA RTX 5060 (8GB) 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.
| NVIDIA RTX 3090 (24GB) | NVIDIA RTX 5060 (8GB) | |
|---|---|---|
| Memory | 24.0 GB | 8.0 GB |
| Bandwidth | 936 GB/s | 448 GB/s |
| Price (approx) | $1,099 | $449 |
| LLMs it runs | 47 of 64 | 22 of 64 |
| Best model it runs | Qwen3.8 27B · 35–53 tok/s | Gemma 4 12B · 55–83 tok/s |
The NVIDIA RTX 3090 (24GB) runs 25 more of the tracked models (47 vs 22), thanks to its 24.0 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.