Token savings versus AI credits
Quick answer
Characters, estimated tokens, the tokens a provider counts and the credits or limits of your plan are four different numbers. Typo shows the first exactly and the second as an estimate. The other two only your provider knows, and some products, like Lovable, bill by work rather than by length.
1. Characters
Your browser and the Typo app can count characters before and after without contacting anyone. This is the only exact number available on this site.
2. Estimated tokens
A simple formula based on characters, words and punctuation gives an approximate token count. Languages, code and punctuation vary per tokenizer, so we always label it as an estimate.
3. Provider tokens
A provider counts tokens with its own tokenizer for the exact model, and the total includes earlier messages, files, cached content, tools and the answer. Some providers offer a counting tool for developers. This site cannot see any of it.
4. Credits, limits and plans
Many AI products count messages, credits or a usage allowance instead of raw tokens. Lovable bills by the work a request causes. Chat apps may have limits that depend on the model and the length of the conversation. A shorter prompt can make little difference there.
A fair comparison
Compare two attempts at the same task: total input, output, caching and tool use, and whether the final answer was accepted. A prompt that saves 20 tokens but causes a second full request costs more overall. That is why we never publish money saved based on characters alone.
What Typo shows
Exact characters and estimated tokens, per change, in the app log and in the Platform Lab. Nothing more, because nothing more can be measured honestly from your computer.
Official sources
Reviewed 8 November 2026. Links go to each provider's own documentation. Typo is not affiliated with any of these companies, and provider terms can change.