The AI in Pairlens is not tied to one company’s model. You choose who provides it, and there are two ways to do that. They are not exclusive: you can have both and switch.
Bring your own key
Free, always, and never gated. If you already have an account with an AI provider, you have an API key: a string from their dashboard that lets a program use your account, billed to you directly.
Install that provider’s plugin from the Plugin Store, paste the key, and the assistant uses it, research included.
Inference. Anthropic, OpenAI, Groq, OpenRouter.
Web search. Tavily, Exa. Without one of these, research reports still work but are built from market data alone.
Your key is stored the same way an exchange key is: in the OS keychain on desktop, in the encrypted credential vault in a browser. Requests go from your machine straight to the provider. Nothing routes through a Pairlens server.
This is the option for anyone who already has an API key, wants a specific model, or wants zero third parties in the path.
Pairlens Intelligence
A subscription for people who would rather not manage API keys at all. Pairlens relays the requests, and you get a monthly budget of credits.
| Plan | Price | Monthly credits |
|---|---|---|
| Intelligence Pro | $19/mo | 13,000 |
| Intelligence Max | $99/mo | 70,000 |
One credit is $0.001 of underlying AI cost, so 13,000 credits is $13 of usage. Chat, research and web search all draw from the same budget, and a web search costs a flat 10 credits plus whatever the request itself uses.
For a sense of scale: ordinary questions cost a few credits each. A full research report is the expensive one, at tens to low hundreds.
Credits reset every billing cycle and do not roll over.
Extra credits
Max subscribers can buy one-time top-up packs that stack on the monthly budget: $10 for 5,000 credits, $20 for 10,000, $50 for 25,000, $100 for 50,000. Pack credits are spent before your monthly budget and expire 30 days after purchase, so an actively used pack forfeits little or nothing.
Prices exclude tax, which is calculated at checkout based on your location.
Manage everything in Settings → Intelligence: current plan, credits used against granted, reset date, checkout, and the billing portal.
Which to choose
| You | Want |
|---|---|
| Already have an OpenAI or Anthropic key | Bring your own key |
| Want a specific model or provider | Bring your own key |
| Run standalone with no App Server | Bring your own key |
| Would rather not think about API keys at all | Intelligence Pro |
| Generate research reports several times a day | Intelligence Max |
Running out
If you exhaust your credits mid-cycle, hosted AI stops until the reset or until you top up. It fails with a clear message rather than degrading silently. A bring-your-own-key provider keeps working regardless, because it is billed by your provider, not by us.
Self-hosted and standalone
With no App Server configured, the terminal runs standalone: auth off, cloud panels hidden, local persistence only. AI still works through bring-your-own-key plugins, with calls going directly from the terminal to your chosen provider. See self-hosting.
What is never sent
No AI request, hosted or otherwise, includes your exchange API keys or wallet private keys. Those live in the OS keychain or your encrypted vault and are read only by the connector signing an order.