AMD Instinct MI250X meets the minimum VRAM requirement for Q4 inference of microsoft/VibeVoice-1.5B. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
AMD Instinct MI250X can run microsoft/VibeVoice-1.5B with Q4 quantization. At approximately 471 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 125GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 3GB | 128GB | 471.47 tok/s | ✅ Fits comfortably |
| Q8 | 5GB | 128GB | 313.61 tok/s | ✅ Fits comfortably |
| FP16 | 11GB | 128GB | 197.31 tok/s | ✅ Fits comfortably |