AMD Instinct MI250X 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.
AMD Instinct MI250X can run petals-team/StableBeluga2 with Q4 quantization. At approximately 510 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 124GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 4GB | 128GB | 509.76 tok/s | ✅ Fits comfortably |
| Q8 | 7GB | 128GB | 345.07 tok/s | ✅ Fits comfortably |
| FP16 | 15GB | 128GB | 175.17 tok/s | ✅ Fits comfortably |