Qwen3.5-9B tops every AI benchmark right now, but that's not how you should pick a model

XDA developers | March 12, 2026, 8:30 a.m. | Original
There's a lot more to a model than just benchmarks.
I fed my Home Assistant logs into a local LLM, and it found problems I'd been ignoring for months Home Assistant can automate things Alexa and Google Home can't even attempt, and it runs on a $35 board