After you install the Python agent for your Large Language Model (LLM) application, Application Real-Time Monitoring Service (ARMS) starts monitoring the application. You can view its token usage on the Token analysis page.
In LLM applications, a token is the basic unit for text processing. It represents the smallest semantic unit of model input and output. A token can be a word, a subword, or a character, depending on the tokenizer that the model uses.
Prerequisites
You have installed an agent for the LLM application. For more information, see Integrate LLM applications or inference services with ARMS.
View token analysis
Log on to the ARMS console, and in the left navigation bar, select .
On the Application List page, select a region in the top navigation bar, and then click the target application name.
In the top navigation bar, click Token analysis.
Panel
Description
Token usage
The total number of tokens consumed by all LLM calls within the specified time range.
Avg tokens per LLM call
The average number of tokens consumed per LLM call.
Avg tokens per request
The average number of tokens consumed per user request.
Token usage/1m
The total number of tokens consumed by all LLM calls per minute.
Avg tokens per LLM call/1m
The average number of tokens consumed per LLM call per minute.
Avg tokens per request/1m
The average number of tokens consumed per user request per minute.
Token Usage Model Ranking (Top5)
The top 5 models by token usage, sorted in descending order.
Token Usage Session Ranking (Top5)
The top 5 sessions by token usage, sorted in descending order.
Token Usage User Ranking (Top5)
The top 5 users by token usage, sorted in descending order.