Token Wise estimates prompt costs and offers optional prompt rewrites

Token Wise is a browser extension that estimates the token cost of a prompt as a user types. It also offers a rewrite action intended to make a prompt shorter or clearer, with a side-by-side comparison of the original and revised text.

Estimates at the prompt box

The vendor lists support for chat interfaces including ChatGPT, Claude, Perplexity, Gemini and several others. The product page describes using published per-token rates for the estimate, with a separate explanation of its calculation method.

An estimate should be interpreted in that context. A subscription’s actual allowance or bill can depend on more than the text visible in one prompt. Review the methodology and compare it with the billing model of the service you use before treating the displayed amount as a final charge.

Local data and optional rewriting

Token Wise says prompt text, live estimates, settings and local history stay on the device while typing. Clicking Optimize sends the prompt to its cloud rewriter, while a local-only mode keeps rewriting offline. These are vendor statements, not an independent privacy audit.

The documented rewrite options include Balanced, Shorter, Clearer and More specific. Check that a shorter rewrite preserves the instructions and context the task actually needs. Saving tokens is useful only if the resulting response still meets the requirement.

Optional monitoring has a defined scope

The extension also describes an API-monitoring option that is disabled by default. It observes metadata for direct browser calls to five supported vendor API hosts; server-side calls are outside that visibility.

For a trial, use a nonsensitive prompt, inspect the proposed rewrite and review the extension’s permissions and current privacy page. The documented dashboard and history can help compare estimates over time, but the product does not establish savings on every request or visibility into all AI spending.