Janitor AI API key: the short answer
There isn’t one universal “Janitor AI API key.” The credential depends on the connection you choose:
- JLLM: JanitorAI describes JLLM as a model available on its platform; this route does not require you to enter an external provider key. (janitorBench)
- janitor+ Router: JanitorAI says Router connects subscribers to supported third-party models without requiring an outside provider key. Its guide says Router is currently web-only and for janitor+ subscribers; model usage is billed through Router. (janitor+ Router)
- External provider through a proxy: Get the API key from the provider—in the walkthrough below, OpenRouter—and enter it in JanitorAI’s chat API settings. JanitorAI’s proxy quickstart documents this route. (Proxy quickstart)
A provider key is for connecting a model to JanitorAI chat; it is not itself a JanitorAI-issued key or evidence of a general-purpose JanitorAI developer API. The steps below cover the documented OpenRouter proxy route, not arbitrary production APIs or custom endpoints. (Proxy quickstart)
Add an OpenRouter key through JanitorAI’s proxy
JanitorAI’s quickstart documents OpenRouter’s Chat Completions URL and the chat configuration flow. Its example model ID is older, so use the current model catalog to choose a model ID instead of copying that example blindly. (Proxy quickstart)
- Create an OpenRouter account and key. Sign up or log in, open the API Keys page, create a key, and copy it when shown. OpenRouter treats API keys as sensitive credentials; keep the key private. (OpenRouter install guide)
- Find a model and copy its current ID. Browse OpenRouter’s model catalog and copy the model slug exactly. For example, OpenRouter lists
deepseek/deepseek-chat-v3-0324as DeepSeek V3 0324. This is a paid listing; check its current pricing and your available OpenRouter balance before using it. Model availability and pricing can change. (OpenRouter quickstart) - Open JanitorAI’s configuration. Start or open a chat, click the gear icon at the top right, then go to API Settings and choose Add Configuration. (Proxy quickstart)
- Enter the proxy details. Use a configuration name of your choice, paste the exact model ID from OpenRouter, and enter this URL in Proxy URL:
https://openrouter.ai/api/v1/chat/completions. Paste your OpenRouter key into API Key without extra spaces. JanitorAI’s quickstart documents this URL and these configuration fields. (Proxy quickstart) - Save, activate, and test. Add the configuration, save settings, and refresh JanitorAI. Reopen the configuration, make sure it is active, and click Test. Then send a short chat message. A successful check means the test does not return an error and the chat produces a reply; JanitorAI’s guide does not specify a particular success-message label. (Proxy quickstart)
Access and billing: The DeepSeek V3 0324 listing above is priced per token, so check OpenRouter’s current model page and your available account balance before choosing it. (OpenRouter model listing)
If the connection fails
First confirm the configuration is active, save settings, refresh JanitorAI, and run Test again. If OpenRouter reports “No endpoints found for model name,” JanitorAI’s troubleshooting guide says the model may no longer exist or be available; return to OpenRouter’s current catalog and copy a listed model ID exactly. (JanitorAI OpenRouter error guide)
If you expose your key publicly, revoke or replace it using your provider’s account controls. OpenRouter advises treating API keys as sensitive. (OpenRouter authentication guide)
Bottom line: For a JanitorAI proxy connection, the external provider issues the key. For janitor+ Router, JanitorAI says no outside provider key is needed. Before relying on an external production model, confirm its current OpenRouter model ID, price, and availability, then test the JanitorAI configuration. (janitor+ Router)
Sources
TL;DR Quickstart Proxy Instructions | janitorai
OpenRouter Error Guide | janitorai