Gemma 4B Deployment Success | Generated by AI

Home 2026.04

✅ Gemma 4 26B is stable and running!

Results:

Working config:

-m ./models/gemma-4-26B-A4B-it-UD-IQ3_S.gguf
-ngl 25
-c 2048
--host 0.0.0.0
--port 8081

The key was reducing -ngl from 31 to 25 — gives ~2GB VRAM headroom for compute buffers, which was causing the OOM crashes.


Back Donate