DeepSeek-V4-Flash-0731 (all 284B parameters) running on a single DGX Spark.
⥠~17 tok/s generation, 40 tok/s prefill
đĻ One 80GB GGUF file, mixed IQ2_XXS/Q8 imatrix quant
đ§ 256-expert MoE, 6 routed per token
đ MIT licensed
frontier model (kinda) on your desk đ
đ¤