Seats & Tokens

Home / Calculator

AI coding cost calculator

Enter how you code each week. We estimate the tokens you'd use, price them at the model's list price, and apply each plan's published billing rule. You can see every assumption and edit it. Prices as of Oct 21, 2026.

What will my AI coding cost per month?

How you work
Assumptions (editable)

Presets are rough editorial assumptions, not vendor data: agent-heavy = 20 requests/h × 40k input (70% cached) × 1.5k output; mixed = 12/h × 15k (50% cached) × 800; light = 6/h × 4k (20% cached) × 500. Autocomplete is free on most paid plans and isn't counted. A month is 52/12 weeks.

Enable JavaScript to run the calculator. All prices are on the tool pages.
Tool · planList priceEst. monthly costHow we estimate

Estimates only. Quota-based plans (Claude, Codex, Windsurf) don't publish limits in tokens, so we can't tell whether your usage fits; we show the flat price and say so. Prices come from the snapshot dated 2026-10-21; each row links its source.

How it works

How does the calculator estimate cost?
It converts your weekly sessions and hours into model requests, multiplies by typical input/output tokens per request (editable), and prices those tokens at the chosen model's list price. Each plan then applies its own published billing rule: a dollar allowance (Copilot, JetBrains, Zed, Augment), subscription plus API-priced usage (Cursor for third-party models), or a flat price with an unpublished quota (Claude, Codex, Windsurf).
Why do some plans say "quota not published"?
Vendors such as Anthropic, OpenAI and Cognition describe limits as multiples of another plan or as message ranges, not as tokens or dollars. We show the flat price and the API-equivalent value of your usage, but we can't say for certain whether it fits.
Is autocomplete included?
No. Tab completion is unlimited or free on most paid plans and doesn't draw on AI credits, so the calculator counts only chat and agent requests.

Full details: methodology.