NVIDIA H200 SXM 141GB meets the minimum VRAM requirement for Q4 inference of Qwen/Qwen3-8B-Base. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
NVIDIA H200 SXM 141GB can run Qwen/Qwen3-8B-Base with Q4 quantization. At approximately 703 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 137GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 4GB | 141GB | 702.66 tok/s | ✅ Fits comfortably |
| Q8 | 9GB | 141GB | 506.27 tok/s | ✅ Fits comfortably |
| FP16 | 17GB | 141GB | 276.97 tok/s | ✅ Fits comfortably |