Grape
Download
Help center
AI setup and models3 min readUpdated October 6, 2026

Set up AI: Grape AI or your own key

The two ways to power the AI features, and which plans include which.

AI features are on the Pro and Lifetime plans, and there are two ways to run them. The Sync plan has no AI on either path: it is sync and sharing alone.

Grape AI

Included with the Pro subscription. Nothing to configure, no keys to manage, no separate provider bill. When Pro starts and no other provider is set up, Grape switches its own models on for you in Settings under Models, and chat, transcription, search, images and the study tools all work.

Step by stepOn desktop
1Click the gear in the editor header to open Settings.
2Go to Models. Grape Pro models are already on for chat, transcription, images and indexing.
3Under Providers, Grape is switched on, with no key to paste.
Step by stepOn mobile
1On the Folders screen, tap the gear.
2Under AI, tap Models.
3Under Providers, switch on Grape.
4Its models are on in every list, and the chat answers with Grape Pro Chat Model.

Your own API key

Available on both Pro and Lifetime. Open Settings, go to Models, enable a provider, paste your key and add the models you want from the ones it offers you. Grape supports OpenAI, Anthropic, Google AI Studio, Groq, DeepSeek, xAI, OpenRouter, and any OpenAI compatible endpoint as a custom provider. On desktop you can also run local models through Ollama or LM Studio.

Step by stepOn desktop
1In Settings under Models, scroll to Providers and switch on yours, like OpenAI.
2Paste your key into the field under it.
3Click Add Model under Chat Models and pick one of the models OpenAI offers you.
4It joins the list, switched on and named with its provider.
Step by stepOn mobile
1In Settings under Models, switch on your provider.
2Paste your key into the field that opens under it.
3Tap Add Model under Chat Models.
4Grape lists the models OpenAI offers you. Tap the one you want.
5It joins the list, switched on.

Four kinds of models

The Models tab separates chat models, transcription models, image models and indexing models, because they do different jobs. Chat powers the assistant and writes mini apps, transcription turns audio into text, an image model generates the images the chat asks for and the pictures in a slide deck, and indexing builds the semantic search index. Add and switch on one of each from whichever providers you use.

Changing your mind

Switch providers whenever you like. The notes themselves do not depend on which model you used, so nothing needs converting when you move from one provider to another.

Continue reading
Add your own AI models and providers
Enable a provider, add the models it offers you, or add one by its id or from an OpenAI compatible endpoint.
Run local AI models with Ollama or LM Studio
Keep AI on your machine with a local model on desktop, and what to expect from it.
How note indexing works
What the index is, why semantic search needs it, and what Sync and Regenerate do.
Did this not help? Contact us with the version you are on and what you were trying to do.

Try It While the Guide Is Open

Grape is free to install and needs no account, so you can follow along in the real app.
Download freeCompare the plans