DGX Spark x4
11recorded runs
190.0best tok/s
11sourced
How it compares — 128 GB per card · multi-GPU
All benchmarks Compare DGX Spark x4 →
| GPU | Model | Quant | Backend | Ctx | tok/s | prompt tok/s | VRAM GB | Source |
|---|---|---|---|---|---|---|---|---|
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash-1017 | EXL3 3.5bpw | vllm bot-lab-21's EXL3 build + drowzeys' … | 300000 | 190.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | GLM-5.3-Flash | EXL3 4bpw | tensorfold v0.6.0 | 500000 | 124.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | GLM-5.3-Flash | EXL3 4bpw | vllm knapcio stack | 500000 | 115.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | GLM-5.3-Flash | NVFP4 | vllm knapcio stack | 500000 | 115.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | EXL3 3.5bpw | vllm e47aa780b | 300000 | 82.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash-1017 | MXFP4+MXFP8 | vllm dsv41-feat (commit e47aa780b) | 300000 | 76.3 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | EXL3 3.5bpw | vllm dsv41-feat commit e47aa780b | 300000 | 73.8 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | FP4+FP8 native | vllm vllm nightly-8a728663 dsv41-feat ove… | 300000 | 73.8 ✓ | 1194.0 | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | MXFP4 | vllm DSpark (Checkpoint) | 300000 | 41.0 ✓ | — | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | MXFP4 | vllm e47aa780b | 300000 | 38.0 ✓ | 1194.0 | — | src |
| DGX Spark x4 · 128 GB | DeepSeek-V4.1-Flash | EXL3 | vllm vllm-exl3 | 1000000 | 30.2 ✓ | — | — | src |