llamaperf

LFM2.5 2.6B

on Unknown GPU · custom engine · 128,000 ctx

Sep 9, 2026
Throughput
17.0 t/s gen
Quant
Q4_K_M (GGUF)

Use cases

agentic

Summary

User reports a custom inference engine running on a OnePlus 13, pure CPU. The engine is built from scratch, is 450kb, and supports other model architectures.