llamaperf

Gemma 4 5.1B E2B

on NVIDIA RTX A6000 48GB · llama.cpp

NVIDIA hardware
May 1, 2026
Throughput
16.9 t/s gen
Quant
bf16 (safetensors)

Use cases

text-generation

Summary

User reports bf16 with no quantization at 10.25 GB VRAM and 61 ms TTFT. Source is dev.to Gaurav Vij.