AI Providers
Provider Overview
| Provider | Cost | API Key | Local | Description |
|---|---|---|---|---|
| Distill Cloud | 1 credit/file (100 free) | No | No | Default. 100 credits free, then from €1.99 for 100 filenames |
| Claude | Billed at Anthropic | Yes | No | Anthropic's Claude models, high analysis quality — no Distill credits |
| OpenAI | Billed at OpenAI | Yes | No | GPT-4o and other models — no Distill credits |
| Gemini | Billed at Google | Yes | No | Google's Gemini models — no Distill credits |
| Ollama | Free | No | Yes | Local LLMs (Llama, Mistral, etc.) — fully offline |
| Vertex AI | Directly with Google Cloud | No (service account) | No | Gemini through your own Google Cloud project — no Distill credits |
Note: Claude, OpenAI, Gemini, and Ollama are only available in the DMG edition. The App Store edition uses Distill Cloud exclusively (see Getting Started, App Store vs. DMG Edition).
Pricing & Credits
Distill Cloud is the default provider. On first launch you get 100 credits free — 1 credit = 1 AI-generated filename. You can top up at any time:
- €1.99 — 100 credits
- €4.99 — 400 credits
- €8.99 — 1,000 credits
No subscription, no sign-up. If you provide your own API key from Anthropic, OpenAI, or Google, no Distill credits are spent — you are billed directly at the AI provider. Ollama is always completely free.
Payment channel: the App Store edition uses Apple In-App Purchases (StoreKit2). The DMG edition opens a Stripe checkout page in the browser; the balance updates automatically once the purchase completes. Promo codes can only be redeemed in the DMG edition (Settings → AI → “Redeem Code”, shown while Distill Cloud is the active provider, or from the buy-credits window).
You can check the current balance any time:
- In the status bar via a sparkles icon with counter (orange at 5 credits or fewer)
- Under Settings → AI with the “Buy More” button and — DMG edition only — “Redeem Code”
- When the balance hits zero, a banner with a direct buy button appears in the drop zone
Setup
Open Settings → AI. Select the desired provider from the list. For your own cloud keys, enter the API key — keys are stored securely in the macOS Keychain.
For Ollama, Ollama must be installed locally with a loaded model. Configure the endpoint (default: http://localhost:11434) and the desired model.
Distill Cloud requires no configuration — credits are debited automatically.
Setting up Vertex AI
Vertex AI is Google’s enterprise offering for the Gemini models. Instead of an API key you store a service account key from your own Google Cloud project. Billing goes directly to Google; no Distill credits are used.
You need three things in the Google Cloud console:
- A project with the
aiplatform.googleapis.comAPI enabled - A service account holding the Vertex AI User role (
roles/aiplatform.user) - A JSON key for that service account (“Keys” → “Add key”)
In Distill, open Settings → AI, pick Vertex AI (Google Cloud) and click Choose JSON file. That is all: Distill reads the file, validates the key right away and shows the service account’s email address as confirmation. The project ID is taken from the file — nothing to type by hand. The key is stored in the macOS keychain, not in the settings.
If the file is rejected, Distill says why — for instance because it is not a service account key, or because the private key is damaged. In that case simply download a fresh key in the console.
Region & data processing
The Region field defaults to global, which carries the full model catalogue. To keep processing inside the EU, enter a region — together with a model that exists there: europe-west3 (Frankfurt) answers for gemini-3.5-flash and gemini-2.5-flash, while the other EU regions carry only the 2.5 series.
Region and model have to match, otherwise Distill reports “model is not available in region”. Two things to know: Google has announced the retirement of Gemini 2.5 — the series stops answering on 31 March 2027, so anyone staying in the EU should switch to gemini-3.5-flash before then. And a region’s catalogue changes; if a name is suddenly rejected, try the other model or go back to global.
What a region buys you: the analysis then runs in that data centre, which is not the case with global. Either way your own Google Cloud agreement and data processing terms apply, and Google does not use your content there for training.
Technically, in Vertex mode Distill talks to exactly two addresses: aiplatform.googleapis.com for the analysis — prefixed by the region, so for instance europe-west3-aiplatform.googleapis.com — and oauth2.googleapis.com for signing in. The sign-in step transmits only your service account name and a signature — no document content.
Vertex AI troubleshooting
| Message | Cause and remedy |
|---|---|
| The service account is missing the “Vertex AI User” role | The key is valid but the account may not access Vertex. Add roles/aiplatform.user in the console. |
| Project was not found, or the Vertex AI API is not enabled there | Check the project ID and enable aiplatform.googleapis.com in the console. It takes a moment to take effect. |
| Model is not available in the region | Inside the EU use europe-west3 with gemini-3.5-flash or gemini-2.5-flash; otherwise set the region to global. |
| Google rejected the key | The key was revoked or deleted in the console. Download a new JSON key and load it again. |
| Google rejected the sign-in because of the clock | The system clock is too far off. Switch to automatic time in System Settings; renaming works again without a restart. |
| Your Google quota is exhausted | Check the quota in the Google Cloud console or try again later. |
Switching Providers
You can switch providers at any time. API keys remain in the Keychain, so switching does not require re-entering them. The provider can also be switched programmatically via the REST API.