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/Ornith-1.0-9B-GGUF
| GPU | Speed | VRAM | Basis |
|---|---|---|---|
| NVIDIA B300 | 240.67 tok/s | 288GB | ✓ Measured |
| NVIDIA B200 | 226.2 tok/s | 192GB | ✓ Measured |
| NVIDIA H200 | 222.94 tok/s | 141GB | ✓ Measured |
| NVIDIA H100 80GB HBM3 | 216.89 tok/s | 80GB | ✓ Measured |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | 201.69 tok/s | 96GB | ✓ Measured |
| NVIDIA A100 80GB SXM4 | 134.67 tok/s | 80GB | ✓ Measured |
| NVIDIA A100 40GB SXM4 | 126.79 tok/s | 40GB | Est. |
| NVIDIA L40S | 114.57 tok/s | 48GB | ✓ Measured |
| NVIDIA A10G | 73.78 tok/s | 24GB | ✓ Measured |
| NVIDIA L4 | 43.35 tok/s | 24GB | ✓ Measured |
| NVIDIA T4 | 32.93 tok/s | 16GB | ✓ Measured |
NVIDIA B300 is the fastest card we've measured running Ornith-1.0-9B, at 240.67 tok/s.
NVIDIA T4 ($2,299 MSRP) is the cheapest card on our bench that runs Ornith-1.0-9B, at 32.93 tok/s.