Weekly Spend Breakdown (Last 7 Days)

Daily LLM cost and call volume answering “How much money did I spend this week?”

Cost by Feature & Configuration (“Which feature costs the most?”)

Breakdown of cost per feature and per configuration settings, sorted from highest to lowest spend.

Feature	Config Settings	Models	Total Cost	This Week	Share of Spend	Calls	In / Out Tokens

Cost by Metadata (“Which user, tenant, or prompt version costs the most?”)

Spend grouped by each metadata key attached with metadata= or tally.context(...) (top 25 values per key).

Key	Value	Total Cost	Calls

Failure & Retry Analysis (“Which calls fail the most?”)

Tracks HTTP 5xx, 429 rate-limit, and network retry attempts and exhausted failures.

Feature	Function	Failures / Rate	Retries	Last Error

Latency Leaderboard (“Which calls are slow?”)

Average, 95th percentile (P95), and maximum response times per feature.

Feature	Function	Models	Avg Latency	P95	Max	Calls

Search Call Logs

Search by feature name, model, config settings, metadata, input/output preview, or error message. Costs marked ~ are estimated from text length because the response had no token usage.

Timestamp (UTC)	Feature	Model	Status	Cost	Latency	Tokens (In/Out)	Retries	Config / Details