Providers

A company that hosts AI models on its own GPU infrastructure and sells access per token through an API. Instead of running a model yourself, you send prompts to the provider's endpoint and pay for the tokens you use — typically in USD per 1 million tokens.

"Trains" tells you whether the provider may use your prompts and outputs to train its models. "Retention" shows how long they keep your request data. Both come from the provider's published policy; providers that never log your data are marked "Zero retention".

Bring your own key: the provider lets you call models with an API key you already own elsewhere, instead of (or alongside) buying tokens from the provider directly.

Start from the model you need — open its page to see every provider that serves it, sorted by price. Then weigh the data policies from this table: for sensitive workloads a zero-retention provider is usually worth a small premium.