BYOK (Bring Your Own Key)
Starcat AI uses a bring-your-own-key model: after you configure a provider and API key under Settings → AI, requests go to the endpoint you specify. This is the current, shipping configuration path — not a “coming soon” feature. Using AI still requires Pro (see Plan comparison).Benefits
- Quota control: use your own API account and billing
- Clear path: requests go straight to your configured endpoint
- Flexible models: switch Chat / Embedding models per task
Supported providers (examples)
The live list in Settings → AI is authoritative.
Setup steps
- Confirm Pro is purchased and active
- Open Settings → AI
- Add a provider; enter API key / Base URL if needed
- Assign models for summary, chat, embedding, translation, and other tasks
- Use Test connection to confirm it works
Security
- API keys are encrypted with AES-GCM and stored in the macOS Keychain
- Starcat does not upload your key in plaintext to its own business servers
- Keep keys safe and set usage limits in your provider console
Related pages
FAQ
Do I have to use OpenAI?
Do I have to use OpenAI?
No. Compatible endpoints, DeepSeek, OpenRouter, Ollama, and others work — follow the list in Settings.
Are keys uploaded to Starcat servers?
Are keys uploaded to Starcat servers?
Keys stay in the local Keychain; requests go to your configured provider. See Privacy.
RAG still doesn't work after setup?
RAG still doesn't work after setup?
Check that you configured an Embedding model separately — Chat alone is not enough. See AI configuration.