llamaperf

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

on NVIDIA RTX 4080 · Strata

Tone: positiveNVIDIA hardware
Nov 8, 2026
Throughput
95.0 t/s gen · 3000.0 tokens/s prompt processing (prefill)Prompt processing is how fast the input is read, not how long you wait for the first word. How to measure time to first token.
Quant
GSQ RCO IQ3_XXS (GGUF)
System RAM
64 GB
VRAM reported
16 GB

Use cases

creative-writing

Summary

User reports Qwen3.8 Flash-Next at about 95 t/s decode and 3000 t/s prefill on an RTX 4080 16GB with 64GB DDR5 5600. Setup is Strata with a GSQ RCO IQ3_XXS quant, driven by opencode to generate a Three.js motion graphics video rendered with Playwright and ffmpeg. The first render took about 40 minutes and roughly 3 followup prompts, about 2 hours in total. User says it is not comparable to Opus but pretty good.

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.

Everything reported on the NVIDIA RTX 4080 →Every Qwen3.8 report →Estimate Qwen3.8 on this card →