Use case
Stop the $40 retry loop
Cost regressions rarely announce themselves. Puse tracks spend per project in real time and alerts Slack when burn rate spikes above your baseline.
Streaming-aware pricing
Token counts are captured from streamed responses as they flow, priced against a per-model table you control. Mid-stream failures still record partial usage — no guesswork at month-end.
Cost-spike detection
Configure rolling windows and thresholds per project. When hourly spend exceeds baseline by your chosen margin, Puse posts to Slack with model breakdown and recent error context.