DGX Spark vs RTX 3090 x2
Best sourced output tok/s per model (single-stream generation) — engine and version, quantization, context and a source link on every row. Hover a value for the quant.
DGX Spark52 sourced runs · best 86.8 tok/s
RTX 3090 x28 verified runs · best 240.8 tok/s
Specs and evidence
| DGX Spark | RTX 3090 x2 | |
|---|---|---|
| VRAM | 128 GB | 24 GB |
| Sourced runs on the board | 52 | 8 |
| Models measured | 9 | 7 |
| Best output tok/s | 86.8 tok/s | 240.8 tok/s |
| Highest measured peak VRAM | — | 42.9 GB |
| Street price (latest snapshot) | — | — |
Head-to-head
On the 2 models measured on both cards, DGX Spark leads on 1, RTX 3090 x2 on 1. The widest gap is on Qwen3.8-27B: DGX Spark 56.0 tok/s vs RTX 3090 x2 219.8 tok/s (RTX 3090 x2 292% faster).
The measured matrix
| Model | DGX Spark | RTX 3090 x2 |
|---|---|---|
| Qwen3.6-27B | — | 240.8 |
| Qwen3.8-27B | 56.0 | 219.8 |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B | — | 197.0 |
| MiniCPM-V 4.7-35B-A3B | — | 172.2 |
| Qwen 3.8 Flash | 86.8 | — |
| Qwen3.8-Flash-Next | 75.7 | 40.1 |
| Nanbeige4.3-33B | — | 66.2 |
| GLM-5.3-Flash | 64.1 | — |
| CYBER-FROST-3.8 | 60.7 | — |
| Qwen3.8-Flash-Next-hibrid48 | 60.0 | — |
| Qwen3.8-Flash-Next-REAP-256-duo | — | 53.5 |
| NVIDIA-Nemotron-3.5-Super-VL-120B-A12B | 48.0 | — |
| DeepSeek-V4.1-Flash | 47.0 | — |
| Underdog-Saluki-27B | 22.6 | — |