Usage analytics

Path: Dashboard /console (embedded charts)

What it does

Slice your usage and spend by time (day / week / month) × model × channel.

Views

Trend

• Aggregated by day: requests, prompt tokens, completion tokens, quota spend
• Default last 7 days; switchable to 30 days / custom

Model distribution

• Horizontal bars: call count and token share per model
• Top-N list: highest-spend models

Channel distribution

• Where your requests actually landed
• Per-upstream success rate and average latency

Per-token slicing

View usage for a specific token in isolation — useful for separating project / customer cost:

• Pick a token → shows charts scoped to it only
• Compare against the account-wide total

Quota alerts

• Yellow warning on dashboard when balance drops below threshold (default 10%)
• Email alerts in multiple languages can be enabled in Personal

FAQ

Q: Are these numbers accurate? A: They match the per-token sums of Logs byte-for-byte, and align with upstream usage. This is the underlying check for the "no token inflation" commitment.

Q: How far back does data go? A: Detailed logs default to 90 days; aggregated daily data goes further.