Add the provider's key in Settings › API, then add a model that uses it in Settings › Text Model or Transcription Model.
- In the Journal tab, tap the gear at the top left to open Settings, then tap API. With no API yet, it lists the providers; otherwise tap Add API.
- Choose the provider. Where it offers both, choose Use For: Text, Transcription or, for OpenAI, Groq, Qwen and Custom, Text and Transcription. Google's card uses Gemini for text and Cloud Speech-to-Text for transcription, each with its own credentials.
- Tap Get API key to open the provider's page for a key, then enter it. For most providers the endpoint is filled in; Custom takes an HTTPS endpoint that works like OpenAI's.
- Give the API a name and tap Add API. Your credential is saved in your iCloud Keychain.
- In Settings › Text Model or Transcription Model, under Your API, tap Add Model, choose the API and a Model from the list, or Enter Model ID where the provider allows it, and tap Add Model. AideMemo uses it at once. One API can serve several models.
In AideMemo Free, Add Model and the models you add show frosted until 5 friends you invited are paying; you can still add the API itself (AideMemo Free).
The providers. For text: Anthropic, OpenAI, Google (Gemini), DeepSeek, Qwen, ByteDance Doubao, Moonshot AI (Kimi), Zhipu AI, MiniMax, OpenRouter, Groq and SiliconFlow. For transcription: OpenAI, Google (Cloud Speech-to-Text), Qwen, ByteDance Doubao, Groq, ElevenLabs, Deepgram, AssemblyAI, Microsoft Azure, Speechmatics, Soniox, Gladia and MOSI. Custom is for any other.
Deleting it. Swipe the API in Settings › API, tap Delete, then Delete API. Models using it show API unavailable until you edit them to use another API.
What the model receives. AideMemo sends what the model needs from your iPhone or iPad directly to that provider, which handles it under its own terms and your account with it. A text model receives your questions and the memory text needed to answer them, including places with their exact locations and your email; Health and Motion data too, unless you turn off Send to online models in Settings › Health. It is never sent your photos. A transcription service receives the recordings you choose to transcribe. Your provider may charge for use. The privacy policy has the details.
With a question you ask in Chat, a text model also receives your notes in Settings › Memory › About You, as many as it can take (About You). A model you add with your own key never files them in categories.
This page answers
- How do I use my own OpenAI API key?
- Can I use Claude, Gemini or DeepSeek as the model?
- Which API providers does AideMemo support?
- How do I remove an API key?
- What does my own model receive?
Something else? See all help articles, or contact us.