Your multi-provider AI spend, turned into a FOCUS-conformant dataset — so the standard FinOps questions have standard FinOps answers.
What this project's own AI usage would cost at API list prices, what the flat subscriptions actually cost, and the leverage between them. Aggregates only.
View the observatoryThe published list rates usage is valued against, plus a dated change log built straight from the git history of the pricing snapshot.
View the price cardtokencur reads local AI agent logs (Claude Code, Codex CLI, Kimi Code), values the
usage at published API rates as showback — subscriptions don't bill per token —
and normalizes it to FOCUS, the FinOps billing
standard. Run python -m tokencur report for the terminal summary.