OpenRouter

OpenRouter is a unified gateway to large language models. Instead of opening a separate account and billing relationship with each model provider, you get one API key and one endpoint that reaches hundreds of models across many providers, and it handles fallbacks between them automatically.

It is OpenAI-compatible: the endpoint at openrouter.ai/api/v1/chat/completions accepts the same request shape most tools already use, so the official OpenAI SDK works as a drop-in replacement by pointing the base URL at OpenRouter. There are also first-party client and agent SDKs for JavaScript and Python. This makes it a convenient way to compare models or switch between them without rewriting integrations.

It fits developers and technical users who want to experiment across models, compare outputs, or avoid committing to a single provider's stack. Billing is pay-as-you-go per token with no monthly subscription, and the catalogue includes a rotating selection of free models you can call at no cost within rate limits — useful for testing before you spend anything.

To begin, create an account at openrouter.ai, generate an API key in the dashboard, and send a standard chat-completions request; pick a model from the models page, including the free ones, by its identifier.

The honest limits, stated plainly. This is a relay in front of other providers, so anything you send passes through a third party — check its terms and avoid sensitive data where that matters. Free models carry tighter rate limits and can change or be withdrawn, and the exact per-token prices of paid models vary by model and move over time.