Which models you need to download and how big they are
There are two models, and each only needs to be downloaded once:
- the speech recognition model — 181 MB, converts your voice to text;
- the large language model (LLM) — 1 GB, analyses the text and finds the specific values for the inspection fields.
By default the app only lets you download the models over Wi-Fi. If you want to download them over mobile data, go to “Settings” → “Voice models” and turn on the “Download over mobile data” switch. The same screen shows what is already downloaded, and later, if you need to, you can delete the models there when your phone is short of free space.