Usage Statistics

The Usage Statistics page is used to centrally view the call counts and Token consumption of each API Key and each model, helping you verify bills, track business call trends, and locate abnormalities in time. This article introduces the page structure, main indicators, and common analysis methods.

Page Overview

• Filter Conditions: The top provides filters such as API Key, Call Model Type, and Statistical Period, which can quickly locate data for a single Key or all models. Supports custom time intervals and quick buttons for "Last 1 Day / 3 Days / 7 Days".
• Query Button and Clear Button: Click "Query" to update data after setting filter conditions. If you need to return to the default view, click "Clear".

Indicator Cards

A set of core indicators is displayed at the top of the page to facilitate immediate grasp of resource consumption:

• Cumulative Model Calls: Total API calls within the current filter range.
• Service Calls: When integrating multiple services or self-developed models, you can view service-level call counts.
• Internal Infrastructure Calls: Used to distinguish between platform internal calls and business calls to help calculate costs.
• Total Token Count: Sum of Token consumption for all requests.
• Total Input Tokens / Total Output Tokens: Display prompt word and generated content consumption separately, facilitating analysis of scenario characteristics.

Usage Trend Chart

The "Usage Trend" module below supports switching the following views:

1. Request Volume: Display the number of calls for each model by day, which can quickly discover traffic peaks.
2. Total Token Count: Observe the overall Token consumption trend to judge cost changes.
3. Input Token Count / Output Token Count: Split the proportion of input and output, which helps optimize prompt word strategies or limit output length.

The legend can display each model (such as cltech/deepseek-r1-32b, deepseek-r1-distill-llama-70b, etc.). Click to temporarily hide a model to focus on key data.

Common Usage Scenarios

1. Cost Accounting: Estimate corresponding fees by combining Token consumption and pricing strategies.
2. Capacity Planning: Observe changes in request volume during peak periods to evaluate whether API Keys need to be expanded or rate-limited.
3. Abnormality Troubleshooting: If there is a sudden increase in calls or abnormal Tokens, the source of the problem can be quickly located through time filtering and model legends.
4. Multi-Key Management: Select different API Keys to compare calling habits and formulate limits or allocation strategies.

Usage Suggestions

• Regularly download or screenshot data for key time periods to retain audit records.
• Set limits or disable abnormal Keys in conjunction with the "API Key Management" page to prevent uncontrolled calls.
• For Keys that are about to expire or exceed quota, you can switch the time window at the top of the page in advance to simulate trends within different intervals.

If you need to export more detailed call details or access custom monitoring, please contact technical support to obtain open capabilities.