What it is
A browser-side token counter and cost table. OpenAI encodings are counted with the real vocabularies; other model families are estimated from the nearest encoding and shown with an error band.
When to use it
Use it before choosing a model, sizing a prompt, or forecasting a monthly API bill from requests per day and average tokens per request.
Gotchas & limits
Claude, Gemini, Llama and Mistral counts are estimates, and chat APIs add a few tokens per message. Prices are reference list prices that change; confirm them on the provider's page. Reasoning models may bill hidden reasoning tokens as output.