AI providers

Paperless Starfruit bundles no AI of its own: on the API Keys screen you connect the providers you want, and the keys never leave your server.

Cloud providers

Each takes a single API key; models come from a curated catalog. Pick the provider, paste the key, and hit Test. A probe call confirms it’s reachable.

Provider	Kind	Example models
Anthropic	anthropic	Claude Opus 4.5, Sonnet 4.6, Haiku 4.5
OpenAI	openai	GPT-4.1, GPT-4.1 mini, GPT-4o, GPT-4o mini, o4-mini
Google	google	Gemini 2.5 Pro, 2.5 Flash, 2.0 Flash
Mistral	mistral	Mistral Large, Mistral Small, Pixtral Large, Mistral OCR

Local & OpenAI-compatible

For self-hosted or third-party gateways (Ollama, LM Studio, vLLM, OpenRouter), choose OpenAI-compatible: give the endpoint a name, set its base URL (e.g. http://localhost:11434/v1), and optionally an API key (many local endpoints don’t need one). Models are discovered live from the server rather than a fixed catalog, and you can add as many endpoints as you like.

Vision support varies

OCR needs a vision-capable model. PDFs are sent as a file to providers that accept them (Anthropic, OpenAI, Google); generic OpenAI-compatible and Mistral endpoints take images. The model picker flags which catalog models are vision-capable.

How keys are protected

• Encrypted at rest (AES-GCM) with a key derived from ENCRYPTION_KEY, which only your server holds.
• Never returned to the browser: editing a saved key shows a “leave blank to keep current key” placeholder, not the value.
• No vendor in the middle: requests go straight from your container to the provider.

Picking models

Which model handles each task is set on the Processing screen: a language model for extracting title/tags/correspondent/date, and an optional OCR model for re-reading scans. They can be the same model or different ones.