API Keys
Bring-your-own keys for the four AI providers basehead plugs into. Drop a key in the row for the feature you want to use; the rest stay empty.
Where to find it
View → Options Page → API Keys (sidebar, key icon — previously labeled “AI”). Each provider gets one row: masked field, Verify button, status line, and Get Key link out to the provider’s signup page.
What you see depends on your edition
basehead only shows the provider rows your edition can actually use — so you’re not asked for keys whose features you can’t reach:
| Provider | Free | Lite | Standard | Ultra |
|---|---|---|---|---|
| Anthropic (Claude) | ● | ● | ● | ● |
| OpenAI | — | — | ● | ● |
| Google Gemini | — | — | — | ● |
| ElevenLabs | ● | ● | ● | ● |
Existing stored keys aren’t deleted on a downgrade — a hidden row’s key just isn’t shown (and its features are gated anyway).
Which key powers which feature
| Feature | Best provider | Notes |
|---|---|---|
| Ask (AI Agent → Ask) | Anthropic (Claude) | 5 free questions / day on basehead‘s key. Add your own Claude key for unlimited + model picker. |
| Voice Input (mic icons app-wide) | OpenAI | OpenAI key required for “AI Enhanced” mode. Otherwise falls back to offline Vosk. |
| Scene Builder (image → search terms) | Google Gemini | Strongest at native image analysis. Ultra-only feature. |
| Tag Creator — audio analysis | Google Gemini | Required when Audio input is ON — only Gemini reads the actual audio. |
| Tag Creator — text-only (filename / description) | Gemini (if keyed) | If a Gemini key is set, Tag Creator always uses it. No Gemini key? Pick Claude or OpenAI. |
| Text Skills — Cleaner, Language Translator, SQL Commander, Category Matcher | Anthropic (Claude) | Default. OpenAI works too if you’d rather use that key. |
| Generator (text-to-speech / text-to-SFX) | ElevenLabs | The only provider for voice + SFX generation. |
The four providers — and what each key unlocks
Anthropic (Claude)
The general-purpose text/reasoning brain. Powers the Chat / “Ask” assistant, the AI Agent, and the text-based AI Skills — Tag Creator descriptions & keywords, The Cleaner, Language Translator. This is the default for chat and text skills, so if you only add one key, this is the one.
OpenAI
Drives Voice Input (speech-to-text) for voice search dictation in the search bar. Required to turn on the “AI Voice” option — without an OpenAI key basehead falls back to the offline voice engine.
Google Gemini
Powers Audio Upload / audio analysis — the mode where the AI actually listens to the sound file (e.g. Tag Creator’s audio-upload mode) to suggest tags and descriptions. Gemini is the strongest at native audio understanding, so it’s required for the audio-upload skills.
ElevenLabs
Powers the Generator — text-to-speech voice generation and text-to-sound-effect generation.
Anthropic / OpenAI / Gemini are interchangeable as the LLM brain. The AI Skills and Agent let you pick any of the three from their own provider dropdown. The notes above describe each provider’s best fit, but you’re free to point a skill at whichever one you’ve added a key for.
How keys are stored
Keys live encrypted on the local machine and are never sent anywhere except the provider’s own API. They save automatically on focus-loss or when you hit Verify; Verify also pings the provider to confirm the key is live.
FAQ
Do I need all four keys?
Nope! Only the providers behind the features you actually use. The minimum for the chat / text skills is Anthropic; Gemini for audio analysis; OpenAI for voice dictation; ElevenLabs for generation.
Why bring-your-own instead of a baked-in basehead subscription?
You pay the provider directly at their rate, you own the relationship, and you can revoke / rotate the key whenever you want. Nothing routes though our servers.
Are my keys synced anywhere?
No cloud sync. Keys are stored encrypted on the local machine. If you want them on a second machine, paste them in there too.