Mastra releases a Metrics API for querying token usage, cost, latency, and errors

Mastra's platform now offers a Metrics API that queries token usage, model cost, latency, and error rates from production projects.

Developers can access it through the Mastra Client SDK (@mastra/client-js 1.9.0 or later), the Mastra CLI (mastra 1.9.2 or later), or observability endpoints.

The API supports totals, breakdowns, time series, and percentile data, with time ranges and status filters.