Skip to content
AI SYSTEM — Artificial Intelligence
deen

Inference costs

Also: Tokenpreise · KI-Kosten

The running costs of using an AI model – usually per token or per image, minute or request.

With API usage you typically pay separately for input and output tokens; output is usually more expensive. Long documents in context, reasoning models and agents with many steps drive costs up.

Ways to save include appropriate model sizes, caching recurring content, shorter contexts and routing simple requests to cheaper models.

General information, not legal or professional advice.