GPU comparison
NVIDIA RTX 4060 (8GB) vs NVIDIA RTX 5090 (32GB) 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 4060 (8GB) | NVIDIA RTX 5090 (32GB) | |
|---|---|---|
| Memory | 8.0 GB | 32.0 GB |
| Bandwidth | 272 GB/s | 1792 GB/s |
| Price (approx) | $530 | $4,399 |
| LLMs it runs | 22 of 64 | 48 of 64 |
| Best model it runs | Gemma 4 12B · 36–54 tok/s | Qwen3.8 27B · 48–72 tok/s |
The NVIDIA RTX 5090 (32GB) runs 26 more of the tracked models (48 vs 22), thanks to its 32.0 GB of memory. The NVIDIA RTX 5090 (32GB) has more memory bandwidth (1792 GB/s), so it generates tokens faster at the same model and quant.