Aggregate view of tokens, costs, and savings from Prompt Packing.
{% if savings is defined and savings %}Tokens saved: {{ savings.tokens_saved | default(0) }}
{% endif %}Per-request Savings Trace rows appear here once requests have been processed. Each row explains what Prompt Packing did on one request (stages applied, tokens before/after, cache origin).
Cache Ledger splits cache savings by origin (TokenPak cache vs. provider cache). Populates once cache_origin telemetry is available.