AMD Instinct MI250X meets the minimum VRAM requirement for Q4 inference of unsloth/DeepSeek-R1-Distill-Qwen-32B-bnb-4bit. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
AMD Instinct MI250X can run unsloth/DeepSeek-R1-Distill-Qwen-32B-bnb-4bit with Q4 quantization. At approximately 166 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 112GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 16GB | 128GB | 166.12 tok/s | ✅ Fits comfortably |
| Q8 | 33GB | 128GB | 105.61 tok/s | ✅ Fits comfortably |
| FP16 | 66GB | 128GB | 65.18 tok/s | ✅ Fits comfortably |