Loading content...
Loading GPU benchmarks...
NVIDIA H200 SXM 141GB meets the minimum VRAM requirement for Q4 inference of meta-llama/Llama-3.2-3B. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 2GB | 141GB | 831.65 tok/s | ✅ Fits comfortably |
| Q8 | 3GB | 141GB | 561.04 tok/s | ✅ Fits comfortably |
| FP16 | 6GB | 141GB | 293.43 tok/s | ✅ Fits comfortably |