llamaperf
Sep 18, 2026
Throughput
10.0 t/s gen
Quant
IQ4_XS (GGUF)
System RAM
16 GB
VRAM reported
6 GB

Use cases

multilingual

Summary

User reports Gemma 4 26B-A4B at over 10 t/s on an RTX 3050 6GB with 16GB DDR4 system RAM. Setup uses the Ultra-Uncensored-Heretic IQ4_XS GGUF. User is using it for translation and finds the results decent.