Tracking GPT Token Usage and Costs via Google Sheets

This workflow automates the process of monitoring and logging token usage and associated costs for language model interactions within a chat environment using n8n, LangChain, and Google Sheets. It starts by receiving a chat message through a webhook and triggers a LangChain AI agent configured to interact with a language model such as OpenAI’s GPT.

The AI agent processes the input, and an embedded callback function captures detailed token usage and cost metrics, which are then logged into a designated Google Sheets spreadsheet for analysis and reporting. Additional nodes set workflow, client, and execution metadata, ensuring each log entry is properly tracked and associated with the correct session.

Furthermore, the workflow includes decision nodes to branch logic based on tool usage, and it logs observability data separately for easier monitoring of token consumption and model performance. This setup is particularly useful for managing AI costs, ensuring transparency in token consumption, and maintaining a clear record of AI interactions in complex applications.