| Model | Quant | GPU | Runtime | Avg acc % | Assist % | Assist-mini % | Latency (s) | Tok/s | Source |
|---|
Accuracy and latency come from different tests and hardware; compare within a column's source only. Blank = not reported.
Which local LLM + GPU actually works for Home Assistant voice control? Every row links to its source.
| Model | Quant | GPU | Runtime | Avg acc % | Assist % | Assist-mini % | Latency (s) | Tok/s | Source |
|---|
Accuracy and latency come from different tests and hardware; compare within a column's source only. Blank = not reported.