Call Logs

Feature Overview

Item	Content
Applicable Roles	Model Provider, Model Consumer
Navigation Path	Model Services > My Calls > Call Logs
Page Route	/modelone/monitoring/calls/log/model
Managed Objects	Personal model-call records, Model ID, attribution, results, usage, latency, and details

Beginner Explanation

Call Logs works like an itemized record of model requests. Locate a record by model, Model ID, status, and time, and use attribution to distinguish the member, project, and Key scope associated with the call.

Terminology

Term	Description
Call Time	The time when the request was made and recorded.
Model ID	The platform model identifier used by the request. Use it to distinguish models with similar display names.
Attribution	The member, project, and Key scope to which the call is attributed.
Call Result	The success or failure state of one call.
First Token Time	The time for a text model to return its first token.
Call Details	Model, status, usage, latency, and error summary for one call.

Recommended Operation Order

Set time and model criteria, query logs, verify result, usage, and latency, and then open details for the target record.

Beginner Checklist

Scenario	Do First	Do Not Do Directly
Cannot find a record	Check time range and time zone first	Repeat a model-name query only
A call failed	Open details for the error summary	Repeat the call immediately
Latency is abnormal	Compare adjacent records for the model	Review one latency value only
Preparing to share details	Remove requests, responses, and credentials	Copy complete details

Prerequisites

1. The current account has access to the Call Logs page.
2. The time range, model, model type, or call status to view has been clarified.
3. Use only redacted log information for troubleshooting.

Page Description

The page shows model-call records for the current account. Query by model name, Model ID, type, status, and time range. The list and details show Model ID and Attribution separately from the model display name.

Page screenshots:

Focus on query criteria, call result, usage, latency, and the details entry.

Main Operations

Query Call Logs

1. Go to Model Services > My Calls > Call Logs.
2. Set a time range, enter a model name or model ID, and select model type and call status if needed.
3. Click "Search" and verify call time, model name, Model ID, Attribution, result, and usage. Click "Reset" if the criteria are incorrect.

View Call Details

1. Click "Details" for the target record.
2. Verify model name, Model ID, Attribution, call time, result, usage, latency, and error summary.
3. For troubleshooting, retain only a redacted request identifier and error summary. Do not copy complete requests, responses, or credentials.

Inspect the call details panel to verify the model identifier, request duration, token consumption, and error response details.

Parameter Reference

Field Name	Required	Field Type	Example	Description
Month	Yes	Month selector	2026-07	Controls the statistical month for call logs.
Date Range	Yes	Date range	2026-07-01 to 2026-07-17	Controls the query time range for call logs.
Model	No	Input	Example Model	Filters call logs by model name.
Model ID	No	Input	<MODEL_ID>	Filters call logs by the exact platform model identifier.
Model Type	No	Selector	Text / Video	Filters call logs by model capability type.
Call Status	No	Selector	Success / Failed / Rate Limited	Filters logs by call processing result.
Minimum Input Tokens	No	Number input	0	Sets the lower input-token boundary for the query.
Maximum Input Tokens	No	Number input	1000	Sets the upper input-token boundary for the query.
Call Time	System-generated	Time	2026-07-17 14:30	Shows when a single call occurred.
Attribution	System-generated	Text	Member / Project / Key scope	Shows the ownership scope to which the call is attributed.
Usage	System-generated	Text / tag	Input 100 / Output 40 Tokens	Shows token, free quota, or multimodal input/output usage.
Time Consumed	System-generated	Time	1.2 s	Shows the total time consumed by a single call.
First Token Time	System-generated	Time	0.3 s	Shows the time before the first token is returned.
Failure Type	System-generated	Text	Authentication	Shows the issue category for a failed request.
Error Message	System-generated	Text	Redacted error message	Shows the error summary for a failed request. Redact it before screenshots or external communication.
Actions	No	Action entry	Details	Opens single-call log details.

Pitfalls

• API Key, Token, Prompt, response content, and complete Endpoint may contain sensitive data. Record only sanitized snippets.
• For 429, check rate limits and quota first. For 401, check credentials and permissions. For 5xx, check model service and upstream status.
• Before retrying, verify request parameters, model version, and input size to avoid repeated invalid calls.

Result Validation

Check Item	Success Signal	If Abnormal
Page is accessible	The My Calls - Call Logs page opens, and My Calls > Call Logs is highlighted in the sidebar.	Check account permissions, navigation path, and page loading status.
Call log list loads	The list shows columns such as call time, model name, Model ID, Attribution, model type, call result, and usage.	Refresh the page or retry after adjusting the month and date range.
Filter controls can be selected	After filtering by month, date range, model, model type, or call status, the list refreshes.	Check whether filters are too narrow, and click "Reset" if needed.
Search / Reset works	Search displays matching logs, and Reset clears the filters.	Check network status, page API responses, and account permissions.
Log details can be opened	Clicking Details opens more information about a single call.	Confirm that the record is still within the log retention period.
Field information is consistent	Model ID, Attribution, call status, time consumed, usage, failure type, and error message are consistent with the details page.	Reopen details or expand the time range for cross-checking.

FAQ

Target Log Is Missing

Symptom:

The target request does not appear after you search by time or model.

Possible Causes:

• The time range, model, or status filter excludes the request.
• Another account or Personal Key created the request.

Resolution:

1. Click "Reset" and select the request date.
2. Narrow the list by Model ID and call status.
3. Verify the account that made the call. If the record is still missing, send the administrator the redacted request time and Model ID.

Call Status Is Failed

Symptom:

The target log shows a failed result or an error.

Possible Causes:

• The Personal Key is invalid or lacks call permission.
• The Model ID, request body, or model parameters are incorrect.
• The quota is insufficient or the target model returns an error.

Resolution:

1. Open "Details" and record the redacted error.
2. Verify the Personal Key, Model ID, request format, and quota.
3. Call again after correction. Contact the Model Provider or administrator if the same error continues.

Call Latency Increases

Symptom:

Total latency or first-token latency is much higher than for similar requests.

Possible Causes:

• The input is longer or the output limit is higher.
• The model or provider responds slowly during the selected period.

Resolution:

1. Compare requests with the same Model ID and a similar input size.
2. Review total latency and first-token latency separately.
3. If latency remains high, send the Model Provider the time range and redacted request IDs.

Token Usage Is Higher Than Expected

Symptom:

Input or output token usage is much higher than expected.

Possible Causes:

• The request contains a long message history or attachments.
• The output limit or generation parameters are too high.

Resolution:

1. Review input and output tokens in Details.
2. Shorten the input or reduce the output limit, and call again.
3. If similar requests still differ, send redacted request IDs to the administrator for metering review.

Log Details Contains Sensitive Data

Symptom:

Details shows credentials, a full Endpoint, customer input, or response content.

Possible Causes:

• The call itself contains sensitive data.
• A screenshot or support ticket was not redacted.

Resolution:

1. Stop copying or sharing the content.
2. Mask the Personal Key, authentication header, request body, response body, and full Endpoint.
3. If a credential was exposed, contact the administrator or security owner immediately and rotate it.

Notes

• Do not expose complete requests, response bodies, Keys, accounts, or cost details in documentation, screenshots, or tickets.
• Call log data may be removed after the configured retention period. Confirm the time range first during troubleshooting.
• Error messages are only used for troubleshooting and must be redacted before public communication.

Next Steps

1. Adjust request parameters or calling methods based on the failure type and error message.
2. To troubleshoot a single request, open Details and view redacted log information.
3. To determine whether the issue is a batch anomaly, go back to My Calls > Call Analytics and view aggregated data.