NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 (NVIDIA) is a 550.0B parameters MoE (55.0B active per token) open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 314.5 GB VRAM for Q4 (weights 302.5 GB + KV 11.0 GB + 1 GB overhead) and 1112.0 GB for FP16. About 0 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 98.6 tok/s (estimated); MI300X 192GB ≈ 65.3 tok/s (estimated); H200 141GB SXM5 ≈ 31.9 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.
| Hardware | VRAM | Quant | tok/s | Fits? | tok/s/$ |
|---|---|---|---|---|---|
| B200 192GB SXM | 192 | Q4 | 98.6 est. | No | 0.002 |
| MI300X 192GB | 192 | Q4 | 65.3 est. | No | 0.004 |
| H200 141GB SXM5 | 141 | Q4 | 31.9 est. | No | 0.001 |
| Gaudi 3 128GB | 128 | Q4 | 17.9 est. | No | 0.001 |
| Mac Studio M2 Ultra 192GB | 192 | Q4 | 9.9 est. | No | 0.001 |
| H100 80GB SXM5 | 80 | Q4 | 7.2 est. | No | 0.0 |
| A100 80GB SXM4 | 80 | Q4 | 4.4 est. | No | 0.0 |
| MacBook Pro M4 Max 128GB | 128 | Q4 | 3.0 est. | No | 0.001 |
| Mac Studio M3 Ultra 96GB | 96 | Q4 | 2.5 est. | No | 0.001 |
| MacBook Pro M3 Max 128GB | 128 | Q4 | 2.2 est. | No | 0.001 |