What it does
LazySusan API gives access to chat, image and video models through one key, one SDK and one bill. The listed models include GPT-6, Claude, Gemini, Grok, FLUX, Nano Banana and Veo. Install the SDK with npm install lazysusan or pip install lazysusan. Create an account in the console, add credits and create a key. The key is shown only once. The JavaScript client reads the key from LAZYSUSAN_API_KEY. Chat calls use ls.chat.create. Image generation uses ls.images.generate. Each call names a model, such as anthropic/claude-sonnet-5-5 or google/nano-banana-2.1. Switching models means changing the model name. The rest of the code stays the same. Chat replies include usage cost in credits through reply.usage.costCredits.
Capabilities
- Each developer and service can have its own key, with roles, monthly budgets and allowed models.
- The console shows who uses what.
- A monthly budget can be set on a key, on a teammate covering all their keys, or on the whole account.
- A request that would go over a budget is refused before it runs.
- Before a request runs, LazySusan reserves the most it could cost, then charges only what was used and releases the rest straight away.
- A request that fails costs nothing.
- A request the balance cannot cover is refused with a 402 before anything runs. Auto top-up can be turned on.
- Teammates are added in the console by entering their email addresses with a role. Each person receives a one-click link.
When to use it
- You want to switch between AI models by changing only the model name in your code.
- You want each developer and service to have its own key.
- You want to cap spend per person or per key so a runaway script stops at the limit.