Day 0 Benchmark: Deploying DeepSeek-V4-Flash-DSpark on GPUStack Doubles Throughput

This article is based on a community benchmark contributed by a GPUStack user. DeepSeek-V4-Flash-DSpark enhances DeepSeek-V4-Flash by adding a Speculative Decoding module. Using the same model weights