llamaperf

DeepSeek V4.1 Flash 552B (16B active)

on 2× NVIDIA RTX 3090 · dsx · 262,144 ctx

NVIDIA hardware
Oct 19, 2026
Throughput
26.0 t/s gen · 170.0 tokens/s prompt processing (prefill)Prompt processing measures input prefill speed; it does not tell you the time to first token.
Quant
2.6bit
System RAM
192 GB

Use cases

codingagentic

Summary

User reports DeepSeek V4.1 Flash at 26 t/s generation and 170 t/s prompt processing on 2x RTX 3090s. Setup is the dsx engine with a 2.6-bit quant (2.6127 bpw experts) and Engram tables, at 262,144 context; the 187.6 GB weights plus 203 GB Engram tables ran from a SATA SSD, and loading from an HDD took 90-110 minutes. The build scores KL 0.1972, 82.59% top-1 and perplexity 9.3626 against the FP8/FP4 teacher; the engine does not support images yet.