I’ve tried a bunch of 2B-12B models to use in home assistant to support the voice pipeline and found Gemma E4B worked best. No real benchmarks, just based on vibes from manual testing. Felt it gave a good balance on executing commands correctly and response time.

I have an atom echo, couple HA VPEs, and the Third Reality Voice. I wouldn’t play music off any of them and have just decided to use my other speakers that are connected to HA(my tv soundbar for downstairs and a Sonos Roam for the bathroom). The ThirdR has a pretty good speaker but still not music quality.