Qwen3.8-27B (NVFP4) on vLLM with MTP: the pre-DFlash2 baseline

Before switching to DFlash2, I ran Qwen3.8-27B (NVFP4) on vLLM 0.27.2rc1 with MTP speculative decoding (3 tokens) on the DGX Spark (GB10). This is the baseline the DFlash2 stack has to beat.

Config: 3 speculative tokens · 262,144 max context · 0.45 GPU-memory utilisation · max 4 sequences.

Near-linear scaling from 1 to 2 concurrent streams — the box has clear headroom left at concurrency 2, which is why the follow-up DFlash2 runs are only the floor, not the ceiling.

This is the number to remember: ~31 tok/s single-stream on MTP. The whole point of moving to DFlash2 is to push past it.