Nemotron-3-Nano-30B-A3B needs roughly 31GB of VRAM at our tested precision. A model that doesn't fit doesn't run slowly, it doesn't run. Here's every card from our bench, sorted by measured speed.
Weights: unsloth/Nemotron-3-Nano-30B-A3B-GGUF
| GPU | Speed | VRAM | Basis |
|---|---|---|---|
| NVIDIA B200 | 345.93 tok/s | 192GB | ✓ Measured |
| NVIDIA B300 | 345.31 tok/s | 288GB | ✓ Measured |
| NVIDIA H200 | 327.48 tok/s | 141GB | ✓ Measured |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | 325.43 tok/s | 96GB | ✓ Measured |
| NVIDIA H100 80GB HBM3 | 316.51 tok/s | 80GB | ✓ Measured |
| NVIDIA A100 80GB SXM4 | 204.06 tok/s | 80GB | ✓ Measured |
| NVIDIA A100 40GB SXM4 | 192.44 tok/s | 40GB | Est. |
| NVIDIA L40S | 190.96 tok/s | 48GB | ✓ Measured |
These cards' VRAM is below the requirement at tested precision, published as data, not hidden.
| GPU | VRAM |
|---|---|
| NVIDIA A10G | 24GB |
| NVIDIA L4 | 24GB |
| NVIDIA T4 | 16GB |
Nemotron-3-Nano-30B-A3B needs roughly 31GB of VRAM at our tested precision. 8 of the GPUs on our bench run it; 3 don't have the VRAM for it.
NVIDIA B200 is the fastest card we've measured running Nemotron-3-Nano-30B-A3B, at 345.93 tok/s.
NVIDIA L40S ($7,500 MSRP) is the cheapest card on our bench that runs Nemotron-3-Nano-30B-A3B, at 190.96 tok/s.