llamaperf

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

on AMD RX 9070 · Strata

Tone: positiveAMD hardware
Nov 14, 2026
Throughput
60.0 t/s gen
Quant
IQ2 (GGUF)
System RAM
48 GB

Use cases

rpcreative-writing

Summary

User reports Qwen3.8 Flash-Next at around 60 tokens/sec on an AMD RX 9070. Setup is Strata with an IQ2 quant, on a Ryzen 7 5800X with 48 GB of DDR4 at 2133 MT/s and an NVMe SSD. The user frames the run as a joke about Sam Altman calling consumer RAM inefficient, and says the model is used for an AI waifu.

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 AMD RX 9070

Bars are medians on one scale, 0 to 73.0 tokens/s. Press ▶ to watch text arrive at a row's speed.

Everything reported on the AMD RX 9070 →Every Qwen3.8 report →Estimate Qwen3.8 on this card →