DeepSeek V4 Flash 284B (13B active)

on NVIDIA DGX Spark · vLLM · 162,816 ctx

Sep 7, 2026

Throughput

20.6 t/s gen · 534.9 tokens/s prompt processing (prefill)Prompt processing measures input prefill speed; it does not tell you the time to first token.

Quant

NVFP4

VRAM reported

128 GB

Summary

User reports REAP-pruned DeepSeek-V4-Flash served at 262k context on a single DGX Spark (Ascent GX10). Setup is vLLM. Prefill and generation throughput remain stable from 4K to 162K context. Benchmarks include pp and tg at various context lengths and concurrency levels.

View source ↗