Page cover

Coin Usage Tab

On the Coin Usage Tab page, there are filter date and four graphs displayed, including Total Coin Spend, Monthly Spend, Top API Keys, and Top Models.

Filter Date

The Filter Date is used to filter data based on a specific date range that you select. With this feature, you can focus the graph display and information only on the relevant time period, such as the last week, this month, or a specific range that is manually specified. This is very helpful in analyzing coin usage trends, model activity, or system performance over a certain period of time. In the Filter Date, display the time range to be displayed and click the Select button.

Total Coin Spend

The Total Coin Spend section is displayed as a pie chart that shows the proportion of coins already used (Spent) compared to the coins still available (Remaining) for the Deka LLM service. In this chart, the blue segment representing Spent appears only as a thin line, while the gray Remaining portion dominates almost the entire circle. This chart provides an instant overview of current efficiency and the remaining capacity available for future usage.

Monthly Spend

The Monthly Spend section displays daily coin usage over a specific period, providing a visual representation of how coins are spent on each particular date. In the chart, dates such as July 7, July 11, and July 15 show high levels of model activity, while dates like July 8 reflect very minimal coin consumption. This information can be useful for monitoring system load, planning resource allocation, and understanding your engagement level with the Deka LLM service.

Top API Keys

The Top API Keys chart displays which users or systems, identified by their API keys, have consumed the most coins within the Deka LLM service. In the chart, each bar represents a different API key (e.g., Person1, Person2, Person3), showing their total coin usage. From the data, it's clear that Person2 has the highest usage, significantly exceeding others, while Person1 shows the lowest activity. This chart is valuable for identifying power users, tracking usage accountability, and optimizing access control or quota settings based on user behavior.

Top Models

The Top Models chart displays the most frequently used LLM models along with the amount of coins consumed by each within the Deka LLM service. Each bar represents a specific model, with models like Llama-4-Maverick-17B-128E-instruct-quantized-w4a16 being the most heavily used, followed by Qwen/Qwen1.5-72B-instruct, while models like BAAI/bge-multilingual-gemma2 show significantly lower usage. This chart helps evaluate user preferences, assess the real-world performance of each model, and support cost-efficiency planning based on the coin consumption per model.

Last updated