RoninForge / AI Stack Cost Auditor
Tick your subscriptions, add your API spend, get one monthly total. Everything stays in your browser.
No upload, no account, no bank login, no email scan.
For context, a single chat subscription plus Copilot runs about 30 to 60 dollars a month; heavy agentic-coding stacks on Max plans commonly clear 200 to 500 before API usage is counted.
Common plans, flat monthly. Tick what you pay for. Every price is editable, so set yours if you are on an annual, regional, or legacy plan.
Usage billing varies month to month. Enter last month's actuals; each row tells you where to find the number.
Find it in the Anthropic console, or drop your usage CSV into the Anthropic API cost analyzer to get the month total by key and model, computed in your browser.
If you pay for Claude Code per token rather than through a Max plan, drop your local ~/.claude logs into the Claude Code spend analyzer and use its month total here. If Claude Code runs on your Max subscription, leave this at
0; the plan above already covers it.
Read it off platform.openai.com under Usage. We do not have an analyzer for the OpenAI export yet, so the console number is the source of truth.
Premium requests or AI credits beyond your plan. The Copilot usage CSV analyzer breaks down GitHub's export, and the Copilot credits calculator estimates the burn before the bill lands.
Your AI stack, per month
$0.00
Saved in your browser only, so your stack is still here next month. Reset clears it.
This is a tally of what you entered, nothing more. Subscription prices are fixed until the vendor changes them; API spend moves month to month, so read the yearly figure as a run rate, not a prediction. Nothing you typed left your browser.
The total tells you how much. These free in-browser tools tell you where it went, per provider. Same rule everywhere: nothing you drop in them leaves your browser.
If Claude Code is your biggest line, drop your local ~/.claude logs and see exactly which project and which git branch burned it.
If you bill Anthropic by the token, drop the console's usage CSV and get the spend split by API key, model, and day.
If you are on Copilot premium requests, see what the same usage will cost under GitHub's AI credits model before the switch surprises you.
If you want next month's Copilot number before the bill exists, estimate your credit burn from how you actually work.
If your Copilot overage line surprised you, drop GitHub's usage export and see which requests caused it.
If API usage dominates your total, compare per-token prices across providers and check whether your workload belongs on a cheaper model.
Subscription prices were verified against each provider's official pricing page on 2026-06-12. A few plans whose official pages block automated checks are marked "estimate" and use the last stable third-party figure; confirm and edit those before trusting them. The page never fetches anything after it loads, and nothing you enter is transmitted anywhere.
Most people undercount because the spend lives in two places: flat subscriptions like ChatGPT Plus or Claude Max, and usage-billed APIs that show up on a different statement. Tick your subscriptions above, add last month's API actuals, and the total is your real number. It is usually higher than the figure people carry in their head.
Every provider shows it in their billing console: Anthropic under Console, OpenAI under Usage, GitHub under your Copilot billing page. If you want the breakdown and not just the total, the Anthropic API cost analyzer reads your usage CSV by key and model, and the Claude Code spend analyzer reads your local logs by project and branch. Both run entirely in your browser.
Both cost 20 dollars a month, so the price is a wash and the answer is which model family you actually reach for daily. The question worth asking is different: if you are paying for both plus a coding subscription, that overlap is the first 240 to 480 dollars a year to scrutinize. Run the tally above and see whether one of them earns its line.
There is no reliable public average; vendor surveys are self-selected and most spending data is private. As a rough shape of the market, a single chat subscription plus Copilot lands around 30 to 60 dollars, while heavy agentic-coding users on Max plans plus API calls commonly clear 200 to 500. Your own tally is the only number that matters, which is what this page is for.
This page is a deliberate one-time tally: fast, manual, and nothing leaves your browser. No bank login, no email scan. If you want the same picture maintained continuously from provider billing data instead of retyped monthly, that is what Goei is for. Bank statements work too, if every AI vendor bills the same card.
No. The page is static, the math runs in your browser, and the only storage is your own browser's localStorage so your stack is still here next visit. There is no backend, no account, and nothing you enter is transmitted anywhere.
Prices last verified 2026-06-12, from each provider's official pricing page.
Independent tool, not affiliated with any provider. Subscription prices change; always confirm against the official page. This page totals what you enter and stores it only in your own browser.

You just built this picture by hand. Goei keeps it current automatically from provider billing data, so next month's number finds you instead of waiting for you to retype it. See the dashboard