GPU comparison
NVIDIA RTX 3090 (24GB) vs NVIDIA RTX 4090 (24GB) for local LLMs
How the two stack up for running open-source LLMs locally — memory, bandwidth, price, and how many of the 48 tracked models each one can run.
| NVIDIA RTX 3090 (24GB) | NVIDIA RTX 4090 (24GB) | |
|---|---|---|
| Memory | 24.0 GB | 24.0 GB |
| Bandwidth | 936 GB/s | 1008 GB/s |
| Price (approx) | $1,099 | $2,999 |
| LLMs it runs | 34 of 48 | 34 of 48 |
| Best model it runs | Gemma 4 31B · 39–58 tok/s | Gemma 4 31B · 41–62 tok/s |
The NVIDIA RTX 3090 (24GB) and NVIDIA RTX 4090 (24GB) run the same set of tracked models; pick on price and ecosystem. The NVIDIA RTX 4090 (24GB) has more memory bandwidth (1008 GB/s), so it generates tokens faster at the same model and quant.