RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can DeepSeek V4 Flash DSpark run on the DGX Spark?

Yes. At 8K context the best fit is Q2_K, using about 108 GB of the 109 GB usable on the DGX Spark. Estimated speed ceiling: ~27 tokens/sec.

Newer compressed-attention design: long-context figures are an upper bound.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)107 GB Tight108 GB Tight110 GB No119 GB No27
3-bit (Q3_K_M)139 GB No140 GB No142 GB No151 GB No—
4-bit (Q4_K_M)174 GB No175 GB No177 GB No186 GB No—
5-bit (Q5_K_M)203 GB No203 GB No206 GB No215 GB No—
6-bit (Q6_K)233 GB No234 GB No236 GB No245 GB No—
8-bit (Q8_0)302 GB No302 GB No305 GB No314 GB No—
16-bit (F16/BF16)568 GB No568 GB No570 GB No579 GB No—

• real GGUF file size. Speed is a bandwidth ceiling (273 GB/s), not a benchmark. Methodology.

More

DeepSeek V4 Flash DSpark on every GPU and Mac · Every model that runs on the DGX Spark · Other models that fit 64 GB VRAM · Try other settings in the calculator

Data: Hugging Face Hub (deepseek-ai/DeepSeek-V4-Flash-DSpark), NVIDIA specs. Data as of 2026-11-17.