Calcis — LLM Cost Estimator for VS Code

See what your LLM API calls will cost before you ship.

Calcis estimates token counts and costs for prompt files directly in your editor. The cost appears in the status bar and updates on every save.

Features

Status bar cost display — shows estimated cost and token count for the active file
Auto-estimate on save — configurable, toggle in settings
All major models — GPT-5.4, Claude Opus 4.7, Gemini 2.5, and 25+ more
Powered by calcis.dev — uses the same estimation engine as the web app and GitHub Action

Setup

Sign up at calcis.dev
Subscribe to the Pro tier or above
Go to your dashboard and create an API key
Open VS Code settings and set calcis.apiKey to your key

Configuration

Setting	Default	Description
`calcis.apiKey`	`""`	Your Calcis API key (starts with `calc_`)
`calcis.model`	`claude-sonnet-4-6`	Default LLM model to estimate costs for
`calcis.autoEstimate`	`true`	Automatically estimate costs when a file is saved

Workspace-level config (`.calcis.json`)

Drop a .calcis.json file at your workspace root to pin project- specific settings that every contributor on the repo will pick up:

{
  "model": "claude-opus-4-7",
  "autoEstimate": true
}

Values in .calcis.json override VS Code settings. The extension watches the file, so edits take effect immediately without a reload. JSON schema auto-completion is contributed, so VS Code suggests the known keys and flags typos.

apiKey is also accepted for parity, but prefer keeping secrets in your personal VS Code settings rather than committing them to the repo.

Usage

Manual estimate

Open a file containing a prompt
Optionally select the text you want to estimate
Run Calcis: Estimate Cost from the command palette (Ctrl+Shift+P)
See the result in the status bar

Auto estimate

When calcis.autoEstimate is enabled, the extension automatically estimates costs every time you save a file. The result appears in the status bar.

Status bar

The status bar shows:

Cost estimate and token count (e.g. "$0.0084 (1,247 tokens)")
Click the status bar item for a detailed breakdown in the output panel

Detailed breakdown

Click the status bar item or run Calcis: Show Details to see:

Input and output token counts
Per-token costs
Total estimated cost
Model and confidence level

Supported models

All models listed on the Calcis models page are supported, including:

Claude (Opus 4.7, Sonnet 4.6, Haiku 4.5)
GPT (GPT-5.4, GPT-5, GPT-4o, o3, o4-mini)
Gemini (2.5 Pro, 2.5 Flash, 3.1 Pro)
And 25+ more

How it works

The extension sends the text content to the Calcis public API which:

Counts input tokens using the model's native tokenizer
Predicts expected output tokens using a trained regression model
Calculates costs using the model's current pricing

All estimation happens on the Calcis server. No LLM calls are made by the extension itself.

Privacy

Your text is sent to the Calcis API for estimation
Calcis does not store your prompt text (only a SHA-256 hash for analytics)
See the Calcis privacy policy for details

Calcis — LLM Cost Estimator

Calcis