The Token & Cost Analyzer answers one question before you spend anything: what will this prompt cost to run? It counts the prompt's input tokens and prices them against any model in the catalogue. It is available on every plan.
Pick a source
A pill above the input switches between two sources.
- Paste text — drop in any text. Nothing is stored.
- My prompt — choose one of your own prompts and its body loads read-only. Drafts count, so you can price a prompt before you ever publish it.
Two shortcuts land you straight in the second source with a prompt already selected: Price across models on your own prompt's detail page, and the same item in the row menu on My Prompts. Only your own prompts are listed.
Read the estimate
The headline is the estimated cost of one API call to the model you picked, with the input-token count under it. A chip says how that count was produced:
- Exact tokenizer — OpenAI families, counted in your browser.
- Provider count — counted by Anthropic's own token API.
- Heuristic estimate — roughly 4 characters per token, for models that publish no tokenizer.
Set Expected output in the right rail (Short, Medium, Long, or a custom number) to fold output cost into the total. A model with no published price reads "Cost unavailable" rather than a misleading zero. Inspect opens the per-token breakdown where one is available.
Compare models
As soon as there is text, a comparison table prices the same body across several models at once, cheapest first, with a multiple against the cheapest row. Models that share a tokenizer are counted once rather than repeatedly.
It leaves no mark on your prompt
Pricing is a read, not a run:
- It does not reveal the prompt or use a premium unlock.
- It does not spend Playground tokens, and records no run.
- It does not count as a view or a use, so your stats stay clean.