EcoMeter AI — Resource Tracker
Track token usage, cost, and environmental impact across your AI conversations — automatically.
EcoMeter AI is a Manifest V3 Chrome extension with a side-panel UI that tracks token usage, cost, and water impact in real time across Claude, ChatGPT, Gemini, Grok, Mistral, Perplexity, Copilot, Poe, and DeepSeek. No account required — everything runs locally in your browser.
It also feeds Legerly's Subscription Auditor. With opt-in, local, lifetime usage tracking enabled, the extension accumulates message counts and token volume per platform/model (no message content). The ⤓ Export for Auditor button downloads this as ecometer-usage.json. The Auditor (audit.html) reads that file client-side via FileReader — never uploaded anywhere — and pre-fills its quiz with measured usage instead of self-reported estimates, so its recommendation engine works off real volume data.
Why Legerly
AI runs on asymmetric information: providers know exactly what each conversation costs — in dollars, energy, and water — and structure things so you don't. That gap is a market failure, and it's helping AI concentrate wealth rather than spread it. Legerly builds privacy-first, ad-free tools that hand that information back to the people using AI, so the benefit flows to them. EcoMeter AI is one of those tools; the full set (AI Price Tracker, the AI Clock, Subscription Auditor, AI Transparency Index) lives at legerlyai.com.
What else is in this repo
This repo holds two products. EcoMeter AI (extension/) is the one this README is about; the rest of the tree is the Legerly website, four self-contained static pages served straight to GitHub Pages from main:
| Page | What it does |
|---|
pricing.html | AI Price Tracker — per-token rates, subscriptions, free tiers, plus what a plan is actually worth |
audit.html | Subscription Auditor — which plan fits your real usage, and the student/teacher routes that genuinely exist |
ai-clock.html | The AI Clock — modelled global AI usage and footprint counters (a projection, re-anchored quarterly) |
transparency-index.html | AI Transparency Index — how openly each provider discloses its environmental impact |
They share extension/prices.json as the single source of truth for prices, which is what ties the extension to the site: EcoMeter measures your real local usage, the Auditor prices it.
Two docs carry the context: PROJECT-CONTEXT.md is the architecture and how-to-ship deep dive, and FRESHNESS.md lists everything in here that goes stale — prices, plan caps, student offers, clock anchors, DOM selectors, store state — with the check that proves each one.
Seven guard scripts, four of them in CI. Run them after touching data:
node scripts/check-prices.js && node scripts/check-auditor.js && node scripts/test-auditor.js && node scripts/check-clock.js && node scripts/test-cost-model.js && node scripts/calibrate-tokenizer.js && node scripts/validate-site.js
Each exists because something was wrong and nothing caught it. They have found a 2× billing error, a 51× image-token error, a false accuracy claim, prices drifting between six files, and a recommendation that sent image users to a tier that throttles image generation.
Features
- Token counting — exact where a real tokenizer applies (bundled cl100k / o200k, or an opt-in provider count API); ±11% on the character-ratio estimator, which is a measured figure, not a guess. The panel labels which method produced each number.
- Cost estimation — live per-message and session totals across all major models
- Water footprint — conservative and full-scope estimates per conversation
- Multi-platform — Claude, ChatGPT, Gemini, Grok, Mistral, Perplexity, DeepSeek, Copilot, Poe
- Optional API keys — add an Anthropic key for exact Claude counts, or a Google key for exact Gemini counts. Both are off by default, each is activated only by adding that key, and each sends only your own message text to that provider's token-count endpoint. Everything works without either.
- Privacy first — nothing leaves your browser except the two opt-in token-count calls above. No telemetry, no analytics, no remote logging. Usage tracking is opt-in, local-only and deletable. The extension self-hosts its fonts; the website still loads Google Fonts, which is a known wart we are not extending.
Supported Models
| Provider | Models |
|---|
| Anthropic | Claude Opus 5.5, Opus 5, Opus 4.8 / 4.7 / 4.6, Sonnet 5.5, Sonnet 5, Sonnet 4.6, Haiku 5.5, Haiku 4.5, plus Fable 5 / 5.1 & Mythos 5 / 5.1 |
| Google | Gemini 4 Argon, Gemini 3.8 / 3.6 Flash, 3.5 Flash-Lite, 3.1 Pro, 3.1 Flash-Lite, 2.5 Pro / Flash / Flash-Lite |
| OpenAI | GPT-6 Astra / 6.1 Sol / Sol / Luna, GPT-5.6 Sol / Terra / Luna, GPT-5.5 (+ Pro), GPT-5.4 (+ mini / nano / Pro), GPT-4o (+ mini), GPT-4.1 (+ mini / nano), o3, o4-mini |
| xAI | Grok 4.7, Grok 4.6, Grok 4.5, Grok 4.3, Grok 4.20, Grok 4, Grok 3, Grok 3 Mini |
| Mistral | Large 3, Medium 3.5, Small 4, Codestral |
| DeepSeek | V4 Pro, V4 Flash, V3, R1 |
The full catalog (including advanced/paid frontier models) lives in extension/prices.json — the single source of truth for pricing, shared by the extension and the Legerly website.
Installation
From source (developer mode)
- Clone or download this repo
- Open Chrome and go to
chrome://extensions
- Enable Developer mode (top right toggle)
- Click Load unpacked and select the
extension/ folder
- Click the EcoMeter AI icon in your toolbar to open the side panel
From the Chrome Web Store
Install the published extension directly: EcoMeter AI on the Chrome Web Store.
Project Structure
ecometer-ai/
├── extension/ # The Chrome extension (load this folder in Chrome)
│ ├── manifest.json # MV3 (version is bumped reactively — see scripts/bump-version.js)
│ ├── sidepanel.html
│ ├── sidepanel.js # Side-panel UI + all logic
│ ├── background.js
│ ├── content.js # Scrapes visible chat text per platform
│ ├── prices.json # SHARED source of truth: api / subscriptions / free_tiers
│ ├── water.json # Per-model water intensity tiers
│ ├── privacy-policy.html
│ ├── icons/
│ ├── fonts/ # Self-hosted (no IP leak to Google Fonts)
│ ├── tokenizer_cl100k.js
│ └── tokenizer_o200k.js
├── scripts/
│ ├── update-prices.js # Writes the prices.json api section
│ ├── bump-version.js # Increments manifest + prices.json version (reactive, on collision)
│ └── roll-clock.js # Rolls the AI Clock anchor forward
└── .github/
└── workflows/
├── publish.yml # Weekly: refresh prices → build → upload draft to Chrome Web Store
├── update-prices.yml # Manual: PR-based price refresh
├── release.yml # On push to main → builds zip + GitHub Release
└── roll-clock.yml # Quarterly: re-anchor the AI Clock, open a PR
This repo also hosts the Legerly website (index.html, pricing.html, ai-clock.html, audit.html, transparency-index.html) served by GitHub Pages from main.
Pricing Updates
The weekly publish.yml workflow (Monday) refreshes prices.json as part of building and uploading the extension draft to the Chrome Web Store. A separate update-prices.yml can be run manually to open a PR with just a price refresh. Either way you review and merge — nothing ships without your approval.
To update prices locally:
node scripts/update-prices.js
Note: update-prices.js currently writes hardcoded values and only covers Anthropic, OpenAI, and Google. Other providers' prices in prices.json are maintained by hand.
Contributing
Issues and PRs welcome. For pricing corrections, the source of truth is:
License
MIT © Legerly
