tokenr-ruby

https://github.com/tokenr-co/tokenr-ruby

Track and attribute LLM costs across OpenAI, Anthropic, and other providers. Async by default, never slows down your app. Supports per-request attribution by agent, feature, team, and custom tags.