Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- The new usage and cost endpoints are in beta and require the beta header. The legacy endpoints remain supported.
- Claude Console and Claude Enterprise organizations use different reporting credentials.
- Rebuilding a legacy daily report requires an explicit time bucket, regrouping the new rows and fetching every page.
- Data usually refreshes within four hours and can take up to 24 hours. For final monthly totals, use months that ended at least seven days ago.
Check the organization and reporting credential first
On October 29, Anthropic announced the beta endpoints GET /v1/organizations/analytics/usage and GET /v1/organizations/analytics/cost. Include the header anthropic-beta: analytics-usage-cost-2026-08-24 with each request. The legacy Usage and Cost API and legacy Enterprise report endpoints remain supported, so you can keep the old reporting path for comparison.
Claude Console organizations use an Admin API credential. For Claude Enterprise, the organization needs Analytics API access. Its primary owner then creates a key in claude.ai with the read:analytics scope. Use credentials authorized for the intended organization and keep secrets out of reports, logs and shared spreadsheets.
How do you rebuild a daily report with the new response?
The legacy API organizes results by time bucket. The new API ranks rows by a metric. In each row, starting_at and ending_at identify its period, group holds its dimensions, and usage or cost holds the metrics. Set time_bucket_width=1d explicitly for a daily report and use the same grouping dimensions. Default ranking does not imply chronological order.
To request time order, set order_by=time and order=asc. Fetch every page, then regroup by starting_at and your selected dimensions. Pages from the same query share a snapshot and report the same data_refreshed_at. Compare the same UTC period, filters and metric across both reports.
What should you check when a day or combination is missing?
Start with data_refreshed_at. Usage after that timestamp can be missing or partial. If none of the requested period has data yet, the timestamp can be null and data can be empty. Use the observed timestamp even if several hours have passed since the underlying activity.
Then check the filter values and organization. An unknown identifier or a misspelled model filter may match nothing. A null in group can mean you did not group by that dimension or that the row has no value for it. After confirming the range has refreshed, the filters match and pagination is complete, fill verified empty buckets with zero in the display time series.
Which reporting windows should you compare?
The start of a query is inclusive and the end is exclusive. For daily buckets, align both boundaries to UTC days. The new API allows up to 31 days per request. Split longer reporting periods into adjacent windows so no day is skipped or counted twice.
For repeat comparisons, set the end of the range at or before a data_refreshed_at you have already observed. That keeps the comparison window fixed. Anthropic can revise totals as late events arrive. For final monthly totals, use calendar months that ended at least seven days ago and record the snapshot timestamp and organization scope.
Sources
Frequently Asked Questions
Why does my legacy report show data that the new report does not?
The APIs refresh on different schedules. Check whether data_refreshed_at covers the relevant period in the new report. Then confirm permissions, UTC boundaries and filters before comparing the same metric.
Do I need daily buckets if I only want the total for a period?
If you do not need a breakdown by day or dimension, omit time_bucket_width and group_by. If the range has matching data, the API returns a total row for the requested period.