Can MiMo V2 Flash run on the DGX Spark?
Yes. At 8K context the best fit is Q2_K, using about 106 GB of the 109 GB usable on the DGX Spark. Estimated speed ceiling: ~46 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) • | 106 GB Tight | 106 GB Tight | 110 GB No | 123 GB No | 46 |
| 3-bit (Q3_K_M) • | 138 GB No | 139 GB No | 142 GB No | 156 GB No | — |
| 4-bit (Q4_K_M) • | 175 GB No | 176 GB No | 179 GB No | 193 GB No | — |
| 5-bit (Q5_K_M) • | 205 GB No | 206 GB No | 209 GB No | 223 GB No | — |
| 6-bit (Q6_K) • | 237 GB No | 238 GB No | 241 GB No | 255 GB No | — |
| 8-bit (Q8_0) • | 307 GB No | 308 GB No | 311 GB No | 324 GB No | — |
| 16-bit (F16/BF16) • | 576 GB No | 577 GB No | 580 GB No | 594 GB No | — |
• real GGUF file size from unsloth/MiMo-V2-Flash-GGUF. Speed is a bandwidth ceiling (273 GB/s), not a benchmark. Methodology.
More
MiMo V2 Flash 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 (XiaomiMiMo/MiMo-V2-Flash), NVIDIA specs. Data as of 2026-11-17.