NVIDIA L4 meets the minimum VRAM requirement for Q4 inference of petals-team/StableBeluga2. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
NVIDIA L4 can run petals-team/StableBeluga2 with Q4 quantization. At approximately 61 tokens/second, you can expect Good speed - acceptable for interactive use.
You have 20GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 4GB | 24GB | 60.59 tok/s | ✅ Fits comfortably |
| Q8 | 7GB | 24GB | 38.39 tok/s | ✅ Fits comfortably |
| FP16 | 15GB | 24GB | 23.10 tok/s | ✅ Fits comfortably |