Does this tool use a real tokenizer?

Yes. For supported model families, this page loads an actual tokenizer in the browser and shows the real token splits for the text you enter. API-side wrappers can still add a small amount of overhead depending on the provider.

Are model prices hardcoded?

No. The page tries to fetch a live pricing catalog at runtime, then lets you override input and output prices manually if you want to model a different rate card or reseller price.

Is my prompt uploaded to a server?

No. Your prompt stays in your browser. The only network requests are for the live model catalog and tokenizer scripts needed to perform local tokenization.

Can I estimate output cost too?

Yes. Set an expected output token count and the page will estimate input cost, output cost, and total request cost using the selected model pricing.

LLM Token Counter & Cost Estimator

Inspect real tokenizer splits, count prompt tokens, and estimate API spend with live presets for OpenAI, Claude, Gemini, and Mistral-family models. Your text stays in the browser.

Real tokenizers Live model presets No prompt upload

What It Shows

Real token splits

The page loads a tokenizer for the selected model family and shows how your prompt is segmented token by token.

Pricing

Live presets first

Model prices try to load from a maintained public catalog at runtime, then remain fully editable for manual overrides.

Estimator

Input + output cost

Estimate single-request spend or multiply by request volume for budgeting, prompt testing, and rough monthly planning.

Model Preset

Provider family

Filter models

Live preset

Presets are refreshed from a live model catalog when available.

Loading live pricing presets…

Tokenizer family

Context window

Preset source

Waiting for live catalog…

Cost Inputs

Input price per 1M tokens (USD)

Output price per 1M tokens (USD)

Expected output tokens

Request count

Heads up: pricing can differ across direct APIs, resellers, batch routes, cached input tiers, or special preview models. Use the preset as a starting point, then override the price fields if your rate card differs.

Prompt

Main input

Optional extra context or system prompt

Extra prompt text

This is counted separately, then added to the main input total.

Chars 0 Words 0 Lines 0 Tokenizer Not loaded

Estimate

Input tokens

Main input plus optional extra prompt text.

Input cost

$0.000000

Per request before output tokens.

Output cost

$0.000000

Based on your expected output token count.

Total cost

$0.000000

For 1 request.

Type a prompt to start tokenizing.

No breakdown yet Enter some text

Tokenization

Token chips will appear here once the selected tokenizer has processed your prompt.

Live by default

The page tries to pull a fresh pricing catalog at runtime instead of shipping a fixed model table that goes stale after the next release cycle.

Browser-only prompts

Your prompt text is never sent to Utilium. Only the pricing catalog and tokenizer code are fetched over the network.

Editable rate card

Every pricing field stays editable, so you can model reseller pricing, cached-input discounts, or future price sheets manually.