Calculator
AI Coding Cost Calculator
Estimate AI coding tool cost per task, per day and per month.
No login. No AI API calls. Runs locally in your browser.
Common scenarios
Bug fix
Small context, 2-4 agent turns, usually the cheapest workflow.
Feature
Several files, tests and review loops. Cost depends on scope control.
Refactor
Higher risk of large context reads and repeated verification cycles.
Claude Code vs Codex vs Cursor
| Tool | Works well for | Cost note |
|---|---|---|
| Claude Code | Best for deep repo context | Can get expensive in long sessions |
| Codex | Good for delegated coding and reviews | Cost varies by model and task depth |
| Cursor | Fast IDE workflow | Subscription plus model usage assumptions |
Token saving tips
- Start with a narrow task and explicit acceptance criteria.
- Limit context to the files needed for the current patch.
- Avoid repeated reads of large logs, lockfiles and generated files.
- Ask for smaller patches instead of one broad rewrite.
- Track cost per accepted change, not only raw token usage.
Want to measure cost per applied patch instead of raw token usage? Try TokenPatch.
Pricing assumptions
Planning estimates only. Verify provider pricing before budgeting.
| Model | Input / 1M | Output / 1M | Last updated |
|---|---|---|---|
| Claude Sonnet 4.6 Anthropic | $3 | $15 | 2026-05-24 |
| GPT-5 / GPT-5 Codex OpenAI | $1.25 | $10 | 2026-05-24 |
| GPT-5 mini / Codex mini OpenAI | $0.25 | $2 | 2026-05-24 |
FAQ
Is this an exact bill predictor?
No. It is an estimate for AI coding workflows. Provider billing, subscriptions, discounts and limits can vary.
Does the calculator call an AI API?
No. All calculations run locally in your browser and no login is required.
Why do coding agents use more tokens than chat?
They repeatedly consume repository context, command output, diffs, plans and review feedback across many turns.