Ranked #4 in desktop gpus by usable model memory — the one feature that decides which LLMs a device can run.
| Usable model memory | 24 GB of 24 GB dedicated vram |
| Largest model, Q4 quality | 32B |
| Chip | NVIDIA Ampere GA102 |
| Memory speed | 936 GB/s |
| Storage | — |
| Display | — |
| Weight | — |
| Operating system | Card — needs a PC |
| AI software | CUDA |
| Price | Check retailer |
24 GB of usable model memory fits a 32B model at Q4 quality, with room left over for a long conversation. Memory decides local AI — everything else is second.
Specifications: NVIDIA RTX 3090 specs.
ModelFit. Laptops, desktop GPUs and mini PCs compared by one feature first: the memory your model can actually use — then everything else.
Specifications from manufacturer listings. Prices from retailer listings; check the retailer for the current price.