llamaperf

Qwen3.8 125B (6B active) Flash-Next

on 3× NVIDIA RTX Pro 6000 Blackwell · Strata · 160,000 ctx

Tone: positiveNVIDIA hardware
Nov 3, 2026
Throughput
150.0 t/s gen
Quant
6bit (GGUF)
System RAM
256 GB

Summary

User reports Qwen3.8 Flash-Next decoding at 150 t/s with Strata on an RTX Pro 6000 plus two RTX 3090s, up from 70-80 t/s on llama.cpp. Setup is Strata with an Unsloth 6-bit quant at 160k context, on a machine with 256GB DDR4. User also reports DeepSeek V4.1 Flash at 2.25-bit decoding at 70 t/s.

Compare this run

Medians of plain runs: one device, one request, no speculative decoding, the whole model in the device's memory. The report above is one setup, so read it beside them and not as one of them.

Qwen3.8 125B · 6B active at 4-bit by hardware

Bars are medians on one scale, 0 to 232.0 tokens/s. Each dot is one run and opens its report. Press ▶ to watch text arrive at a row's speed.

Models most run on the NVIDIA RTX Pro 6000 Blackwell

Bars are medians on one scale, 0 to 232.0 tokens/s. Each dot is one run and opens its report. Press ▶ to watch text arrive at a row's speed.

Everything reported on the NVIDIA RTX Pro 6000 Blackwell →Every Qwen3.8 report →Estimate Qwen3.8 on this card →