Resource
Token Counter for AI Prompts
Paste a prompt, get an honest token estimate — a range, not a fake-precise number — plus the cost across GPT, Claude, and Gemini.
Estimate Token Usage Before You Run It
Know how many tokens a job will consume before you send it — input plus an assumed response, costed per call and at scale.
Overview
Token usage is input plus output, and output usually costs several times more per token. This loads a longer document with a long assumed response so you can see the full consumption, not just the prompt half. The estimate is a range across the tokenizer's real uncertainty, and the per-1,000-calls line shows where a cheap-looking call turns into a real bill. Usage is a consumption question — distinct from "will it fit", which is the Estimator's job.
How to use this resource
-
Set the response size
Output tokens dominate cost — pick the assumed reply length.
-
Read input plus output
The combined line is the real per-call consumption.
-
Scale it
Multiply by volume with the per-1,000-calls figure.
Why This Works
- Usage is input + output, and the tool costs both, not just the prompt
- Output tokens are weighted at their real, higher price
- The range keeps the estimate honest instead of falsely exact
Best for
- Planning consumption for a batch job
- Including output cost, not just the prompt
- Estimating before committing to a run
Not for
- Checking if it fits the context window — use the Context Window Estimator
- Trimming the prompt itself — use the Prompt Cleaner
FAQ
Why does the token estimate show a range like 855 to 1,045 instead of one exact number?
The range spans the real uncertainty between tokenizers — the same text lands at ~950 on Claude but ~905 on GPT-5 and Gemini Pro in this report, so a single figure would be falsely precise. The Token Counter derives these from character-based heuristics in your browser, not live tokenizer output, so treat the band as a planning window rather than a billed count.
Is the model's reply counted in this cost, or only the tokens in my prompt?
It costs both. The report adds an assumed reply — here a 'Long response (~2,500 tokens)' — on top of the 3,582-character input, then shows a 'Combined per call' line ($0.0401–$0.0406). Since output is priced at $15/1M versus $3/1M input for Sonnet, the assumed response drives most of the cost, which is why the prompt-only number understates the real bill.
How do I turn a cheap-looking per-call cost into a realistic budget for a batch job?
Read the 'Per 1,000 calls' line, not the per-call one — this report turns $0.0401 a call into $40.06–$40.63 per thousand. A single request always looks trivial; that scaled line is where token cost becomes real, so multiply it by your actual volume when you budget rather than eyeballing one request.
The report lists GPT-5, Claude Opus, Sonnet, and Gemini Pro counts — does that tell me whether my prompt fits the context window?
It only counts tokens per model (GPT-5 ~905 · Opus ~950 · Sonnet ~950 · Gemini Pro ~905), which is a consumption figure, not a fit check. The report says so directly and points to the Context Window Estimator for 'will it fit?'. Use these numbers to compare per-model spend, then run the prompt in whichever assistant you chose to see actual usage.
More resources from Token Counter
Resource
Calculate AI API Cost for a Prompt
Turn a prompt into a dollar figure: input cost, output cost, combined per call, and the number that actually matters — cost per 1,000 calls.
Resource
Count Tokens Before Sending to the API
A quick pre-flight check: count a system prompt's tokens and cost before it ships, so the bill and the size hold no surprises.
Resources that pair well
Resource
Estimate Token Budget — Plan Before You Paste
Token budget planning for real workloads: how much of the window a transcript actually consumes, what is left for the answer, and how much headroom remains.
Resource
Message Too Long — the Fix That Doesn't Butcher Content
The "message too long" error has a structural fix: split at paragraph boundaries into sequenced chunks with wait rules, instead of pasting fragments and hoping.
Resource
Prompt Cleanup Examples (Before & After)
A set of before-and-after examples showing exactly what prompt cleanup removes — and what it deliberately leaves alone.
Related tools
Tool
Token Counter
Estimate how many tokens a prompt is and what it costs — honest ranges across GPT, Claude, and Gemini, with per-call and per-1,000-call pricing.
Guides for this resource
Guide
How to count tokens in a prompt before you send it
Counting a prompt's tokens before you send it tells you whether it fits the model, what it will cost, and whether the end might get cut off. Here's how to check and trim.