OpenRouter, Inc

OpenRouter is a single API for AI models from many providers. Instead of signing up with each model vendor, you send requests to one endpoint and choose from more than 600 models for text, images, video, audio and other tasks, served by 80+ providers.

The platform picks a model and provider for each request using live data from its marketplace. Failover is on by default, so a request that fails at one provider can be sent elsewhere without you writing retry logic, and failed attempts are not charged for their model tokens. You can also add your own provider keys for extra capacity.

Teams get one dashboard that shows requests, tokens and spend across models and providers. Admins can set budgets with alerts, choose which providers may see prompts, decide whether they can keep data, and pick EU or US processing regions. Workspaces, audit logs and SSO are available on the higher plans.

It is aimed at developers and companies that build AI features into their products and want to avoid depending on a single model provider. To start, you create an API key and send a first request. The Free plan includes 25+ free models and 50 requests a day. Standard and Business are pay-as-you-go with no subscription and no minimum spend. You buy credits, and each request deducts its cost from your balance. Inference is billed at the provider's list price, and OpenRouter charges a fee when you buy credits: 5.5% on Standard and 8% on Business. Enterprise pricing is set by contract.