How does the live count work?

The meter estimates tokens locally from your text using a character-and-word heuristic (about 1 token per 4 characters), then multiplies by the model's input price. It updates instantly as you type because nothing leaves your browser.

Is the token estimate exact?

It is a close approximation, typically within a few percent. For an exact, billable count use a provider token-counting tool. The live meter is built for fast, cost-aware drafting.

Why include a completion length?

Output tokens are usually 3–5× more expensive than input, so the prompt cost alone understates a call. The completion estimate lets the meter show realistic total cost per request.

Are the prices up to date?

They are editable presets based on published list prices and clearly labelled as estimates. Confirm the current rate in your provider dashboard before budgeting.

Is my prompt sent anywhere?

No. The meter runs entirely in your browser. Nothing you type is uploaded, stored or logged.

Live Prompt Cost Meter

Live prompt cost meter

Write prompts with a price tag in view. As you type, this meter shows a running token count and dollar cost for your selected model and expected completion length — so you feel the cost of a verbose system prompt before it ships to production.

How it works

The meter estimates tokens locally from your text, blending a character-based rule (≈1 token per 4 characters) with a word count for stability across short and long inputs. It then prices the call:

prompt_cost = (prompt_tokens / 1,000,000) × input_price
output_cost = (expected_completion / 1,000,000) × output_price
total       = prompt_cost + output_cost

Because output is priced several times higher than input, the completion-length field often dominates the total — a useful reminder that capping max_tokens is the cheapest optimization.

Tips

Trim the system prompt. A bloated system message is paid on every single call; the meter makes that cost visible.
Set a realistic completion length. Estimate your typical answer size so the total reflects real usage, not just the prompt.
Verify before launch. For exact, billable counts, confirm with a provider-backed token counter before locking in a budget.