Export Account Logs

GET /api/open/v1/logs/export

Returns every matching log row in a single response instead of paging. The filters are the same as Get account logs; the response is a file in CSV or JSONL format.

Request parameters

Parameter	In	Type	Required	Default	Description
`start_timestamp`	query	integer	No	See below	Start time (inclusive), unix seconds
`end_timestamp`	query	integer	No	See below	End time (inclusive), unix seconds
`type`	query	integer	No	`2`	Log type: `2` for consumption (billed) requests, `7` for final failed requests (not billed). Repeat the parameter or separate values with commas (`?type=2,7`); other values return 400
`model_name`	query	string	No	—	Exact model name; repeat the parameter for multiple values (`?model_name=a&model_name=b`) or separate values with commas
`key_name`	query	string	No	—	Exact Key name; repeat the parameter for multiple values. Commas are not supported because names may contain commas
`key_id`	query	integer	No	—	Key ID, a positive integer; repeat the parameter or separate values with commas. IDs that belong to other users return no rows
`format`	query	string	No	`csv`	Output format: `csv` or `jsonl`; other values return 400

Time range rules:

Both omitted: the last 30 days.
Only start_timestamp: 30 days forward, capped at the current time.
Only end_timestamp: 30 days backward.
The range can be at most 366 days.

Each multi-value parameter accepts at most 50 values; more returns 400 invalid_request. Timestamps must be non-negative integers, otherwise 400. This endpoint does not accept p or page_size.

Request example

curl "https://www.routeapi.ai/api/open/v1/logs/export?start_timestamp=1726408000&end_timestamp=1729000000&key_name=bot&type=2,7&format=csv" \
  -H "Authorization: Bearer $ROUTEAPI_MANAGEMENT_KEY" \
  -o logs.csv

Response

The response body is the exported file, not the usual { success, message, data } JSON. Headers:

Header	Description
`X-Export-Rows`	Number of rows written to the file
`X-Export-Truncated`	`true` when more rows matched than the per-request cap; `false` otherwise
`Cache-Control`	Always `no-store`
`Content-Disposition`	`attachment; filename="logs-<UTC yyyyMMdd-HHmmss>.csv"` or `.jsonl`

Each request returns at most 50,000 rows. When more rows match, the newest 50,000 rows by log ID are returned and X-Export-Truncated is true; split the time range into smaller windows to fetch the rest. Unlike Get account logs, rows are sorted by log ID descending.

CSV format

format=csv returns Content-Type: text/csv; charset=utf-8 with a UTF-8 BOM, followed by a header row and one row per log. Columns are always in this order:

id,created_at,type,key_id,key_name,model_name,prompt_tokens,completion_tokens,total_tokens,amount,request_id,key,use_time,is_stream,first_response_ms,cache_read_tokens,cache_creation_tokens,reasoning_effort,request_path,web_search_calls,audio_input_tokens,audio_output_tokens,billing_source
12345,1729000000,2,12,bot,gpt-4o,1200,300,1500,0.05,20261020120000123456789,sk-abcd**********wxyz,3,true,420,800,0,medium,/v1/chat/completions,0,0,0,wallet

The fields match the items from Get account logs; created_at is unix seconds. Cells that start with =, +, @, a tab, or a carriage return, or that start with a non-numeric -, are prefixed with ' to prevent spreadsheet software from treating them as formulas.

JSONL format

format=jsonl returns Content-Type: application/x-ndjson with one JSON object per line, using the same fields as the items from Get account logs:

{"id":12345,"created_at":1729000000,"type":2,"key_id":12,"key_name":"bot","key":"sk-abcd**********wxyz","model_name":"gpt-4o","prompt_tokens":1200,"completion_tokens":300,"total_tokens":1500,"amount":0.05,"use_time":3,"is_stream":true,"first_response_ms":420,"cache_read_tokens":800,"cache_creation_tokens":0,"reasoning_effort":"medium","request_path":"/v1/chat/completions","web_search_calls":0,"audio_input_tokens":0,"audio_output_tokens":0,"billing_source":"wallet","request_id":"20261020120000123456789"}

Errors

HTTP	code	Scenario
400	`invalid_request`	A timestamp is not numeric or is negative, the range exceeds 366 days, `start_timestamp` is greater than `end_timestamp`, a multi-value parameter has more than 50 values, `key_id` is not a positive integer, `type` is not `2` or `7`, or `format` is not `csv` or `jsonl`
429	—	Per-user query rate limit exceeded (default 30/minute); the body may be empty

Errors are returned as the usual JSON error body, not a file. For other common error codes, see Platform API errors.

Notes

By default only consumption (billing) logs are exported. Add type=2,7 to include final failed requests (not billed) as well.
key_name is the Key’s name at the time of the request; renaming a Key does not rewrite past records.
key is the masked value; it is an empty string when the Key has since been permanently deleted.
Token counts are raw counts, not millions. Amounts use the site display currency; see Get balance.
This endpoint shares the per-user query rate limit (default 30/minute) with Account logs, Usage, Usage by model, and Usage by Key. See Rate limits.