DeepSeek V4 Flash 284B (13B active)
on 4× NVIDIA RTX 4090 · vLLM
Summary
User reports prefill speeds of ~5000 pp/tg for DeepSeek V4 Flash on 4x RTX 4090 with a vLLM fork.
User also mentions Qwen3.8 Flash next with ~7500pp/135tg (MTP).
User compares to llama.cpp and ik_llama, noting vLLM is much faster for prefill.