DeepSeek-V4-Flash-NVFP4 vs NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16
Specs, VRAM requirements and download trends — updated 6 August 2026.
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | DeepSeek-V4-Flash-NVFP4 | NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 |
|---|---|---|
| Parameters | 166.7 B | 560.5 B |
| Architecture | DeepseekV4ForCausalLM | NemotronHForCausalLM |
| Context length | 1,048,576 | 262,144 |
| Licence | mit | other |
| Commercial use | Allowed | Unknown |
| Languages | — | 12 |
| Downloads 30d | 202,991 | 213,329 |
| Downloads all time | 1.9 M | 940 K |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2026-05-18 | 2026-06-03 |
Download trend
Daily snapshots, last 56 days (28 Jul – 21 Sep)
1.3 M203 K
DeepSeek-V4-Flash-NVFP4
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
DeepSeek-V4-Flash-NVFP4 · fp16 @ 1024K ctx
3,386.8 GB / 24 GB
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 · fp16 @ 256K ctx
3,924.2 GB / 24 GB