llamaperf

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

on AMD RX 6750 XT · Strata · 256,000 ctx

Tone: mixedAMD hardware
Nov 1, 2026
Throughput
25.0 t/s gen · 150.0 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
4-bit
KV cache
4-bit
System RAM
32 GB
VRAM reported
12 GB

Use cases

codingagentic

Summary

User reports Qwen3.8 Flash-Next at about 25 t/s generation and about 150 t/s prompt processing on an RX 6750 XT with 32GB DDR4 system RAM. Setup is Strata with a 4-bit quant, 4-bit KV cache and 256k context; the Gyro-S variant is used. User asks why models stop short on large reverse-engineering tasks and whether local Flash-Next is worth it versus a $20 subscription, noting it is slower and less capable than GPT-6 Luna.

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 AMD RX 6750 XT →Every Qwen3.8 report →Estimate Qwen3.8 on this card →