Skip to content
Build Perch
Browse documentation

OpenRouter setup

Short answer: Create an API key at openrouter.ai, paste it into Build Perch’s Integrations window, and you get the live catalog: hundreds of models from OpenAI, Anthropic, Google, and more. OpenRouter bills your account directly per message; Build Perch never sees or marks up that cost.

What is OpenRouter?

OpenRouter is a gateway: one API key gives Build Perch access to the full range of chat, image, speech, and video models OpenRouter lists, across every major lab, without a separate account or key per provider. You pay OpenRouter, OpenRouter pays the upstream provider, and Build Perch never sits between you and that billing relationship.

The catalog is fetched live from OpenRouter’s API (cached for 10 minutes), so a model that launches today is selectable in Build Perch today, no app update required.

How do I set up OpenRouter?

  1. Open the app’s Integrations window (from the model picker’s “API Key Manager” link, or the toolbar) and select OpenRouter under LLM Providers.
  2. Click Get API key to open openrouter.ai/keys in your browser, create a key, and copy it.
  3. Paste the key into the OpenRouter row and save. Build Perch tests the connection and confirms once it’s valid; the key itself is never shown again, only a “Saved” status.
  4. Open the model picker (Connect AI Model, or Switch Model on an existing chat) and pick a model from the OpenRouter list. Each row shows its context window, its rate ($/1M tokens in and out), and whether it supports tool calls.

Good to know

  • You pay OpenRouter per message; there is no subscription. OpenRouter bills your account directly at the rate shown next to each model; Build Perch takes no cut and shows no aggregate spend total, only that per-model rate with a link out to the model’s OpenRouter page.
  • Zero Data Retention is on by default. Every OpenRouter request Build Perch makes (chat, media generation, background summarization) carries a zero-data-retention flag, so OpenRouter routes only to providers that attest they don’t store or train on your prompts. This is enforced by OpenRouter’s routing based on the provider’s stated policy, not a guarantee Build Perch makes on their behalf. Turning it off (Integrations → OpenRouter) makes free-tier models available that don’t offer a zero-retention option, at the cost of that protection.
  • Two more OpenRouter-only toggles live next to the key, both on by default: keeping a chat pinned to one upstream server so repeated context gets cached (cheaper on long chats), and holding that cache for up to an hour instead of the usual ~5 minutes.
  • Free models carry extra caveats. Picking a model priced at $0 shows a warning that free models have strict rate limits and, unlike most paid models, often aren’t covered by a zero-retention guarantee.
  • A model without tool support is flagged. The picker marks chat models that don’t support function calling with a “No Tools” badge, useful to know before you pick one for a mode that relies on file or web tools.
  • Without a key, only local Ollama models are available. OpenRouter is one of three providers; you don’t need it if you’re running local models or an Ollama Cloud subscription instead.