Choose Your Path
Choose the correct edition and decide between local or cloud AI.
Follow the bold button names and numbered steps. Technical names are included only when you need to recognize a model or file inside the app.
Which edition do you have?
| If you obtained Local Waifu from… | Edition | What happens |
|---|---|---|
| The Local Waifu website or official Direct release | DIRECT | Full 7-day trial, then a lifetime license key is required |
| A Store/Unlocked package such as the dedicated itch.io build | STORE | Full access without entering a key inside the app |
| A storefront that resells the app, such as BOOTH | Depends on the package that store serves | Check which build you received with the test below |
A quick way to check is Settings → License:
- The tab is present in the Direct edition.
- The tab is intentionally absent in the Store/Unlocked edition.
Do not install a Direct update over a Store/Unlocked copy merely to obtain a newer version. Keep each edition on its own update channel. If a storefront resells the app, follow that storefront’s download and update instructions, and confirm the edition with the Settings → License test above before changing anything.
Local AI or Cloud AI?
| Choice | Best for | Requirements | Privacy note |
|---|---|---|---|
| Local AI | Private offline-capable chat and no per-message provider bill | Model download, enough RAM and disk space | Chat inference stays on your computer |
| Cloud AI | Smaller computers or access to a provider’s hosted models | Internet and your own provider account/key | Prompt and relevant context go to that provider |
Memories, character data, and the Local Waifu database remain local even when you intentionally choose a cloud model. Cloud image, speech, Telegram, and remote Ollama routes transmit the data needed by those services.
Cloud AI is also never a silent backup. If the local engine is down, or she receives a photo that no local model can see, the message does not go to a saved cloud key unless you turned on Use the cloud when the local model can’t answer in Settings → Performance → Network. That switch is off by default. Chapter 29 explains it.