ai-token-cost-meter

Real-time token cost tracker across multiple AI providers — Claude, OpenAI, Gemini, and more.

Install

npm install ai-token-cost-meter

Quick Start

const { meter } = require('ai-token-cost-meter');

meter.track('claude-sonnet-4-6', 1000, 500);
meter.track('gpt-4o', 2000, 800);
meter.track('gemini-1.5-flash', 5000, 1200);

meter.report();

Output:

┌──────────────────────────────────────────────────────────────────────────────────────┐
│ Model                         │ Provider     │        Input │       Output │        Total │       Cost │
├──────────────────────────────────────────────────────────────────────────────────────┤
│ claude-sonnet-4-6             │ Anthropic    │          1.0K │         500 │         1.5K │   $0.0105 │
│ gpt-4o                        │ OpenAI       │          2.0K │         800 │         2.8K │   $0.0130 │
│ gemini-1.5-flash              │ Google       │          5.0K │         1.2K │         6.2K │   $0.0008 │
├──────────────────────────────────────────────────────────────────────────────────────┤
│ TOTAL                         │              │          8.0K │         2.5K │        10.5K │   $0.0243 │
└──────────────────────────────────────────────────────────────────────────────────────┘

API

meter.track(model, inputTokens, outputTokens)

Track token usage manually.

meter.track('claude-sonnet-4-6', 1500, 800);

meter.trackFromResponse(model, response)

Automatically read token counts from an AI SDK response object. Supports Anthropic, OpenAI, and Gemini response shapes.

// Anthropic
const response = await anthropic.messages.create({ ... });
meter.trackFromResponse('claude-sonnet-4-6', response);

// OpenAI
const response = await openai.chat.completions.create({ ... });
meter.trackFromResponse('gpt-4o', response);

// Google Gemini
const response = await model.generateContent({ ... });
meter.trackFromResponse('gemini-1.5-pro', response);

meter.report()

Print a formatted cost table to the console.

meter.totalCost()

Returns the total cost of the current session as a number.

console.log(`Total spent: $${meter.totalCost()}`);

meter.totalTokens()

Returns total token count across all models.

meter.budget(amount)

Set a spending budget in USD. Logs a warning when exceeded.

meter.budget(1.00); // warn if session exceeds $1.00

meter.onTrack(fn)

Register a callback that fires on every track() call.

meter.onTrack((entry, totalCost) => {
  console.log(`${entry.model}: ${entry.inputTokens} in / ${entry.outputTokens} out — total $${totalCost}`);
});

meter.addModel(id, inputPricePerM, outputPricePerM, provider)

Register a custom model with your own pricing (price per 1M tokens).

meter.addModel('my-fine-tuned-model', 5.00, 20.00, 'MyProvider');
meter.track('my-fine-tuned-model', 1000, 500);

meter.export(format)

Export session data. Supports "json" (default) and "csv".

const json = meter.export('json');
const csv  = meter.export('csv');
require('fs').writeFileSync('costs.csv', csv);

meter.getSession()

Returns raw session data as an object.

meter.getHistory()

Returns the full call-by-call history array.

meter.reset()

Clear all tracked data for the current session.

meter.listModels()

Returns an array of all built-in supported model IDs.

Supported Models

| Provider | Models | |-----------|--------| | Anthropic | claude-opus-4-7, claude-sonnet-4-6, claude-haiku-4-5, and Claude 3.x series | | OpenAI | gpt-4o, gpt-4o-mini, gpt-4-turbo, gpt-4, gpt-3.5-turbo, o1, o1-mini, o3-mini | | Google | gemini-1.5-pro, gemini-1.5-flash, gemini-2.0-flash, gemini-2.0-flash-lite | | Meta | llama-3.3-70b, llama-3.1-8b | | Mistral | mistral-large, mistral-small, mixtral-8x7b |

Custom Instances

For multiple isolated sessions (e.g. per-user tracking):

const { Tracker } = require('ai-token-cost-meter');

const userTracker = new Tracker();
userTracker.track('gpt-4o', 1000, 500);
console.log(userTracker.totalCost());

License

MIT