Skip to content
devprohacks.in

Prices as of 18 Oct 2026 · incl. DeepSeek V4.1 Flash & GLM 5.3

Are you overpaying
for AI coding?

Find out in 10 seconds, without uploading anything. Drop your Claude Code, Codex, OpenCode or Cursor usage and get a savings range, the plan that fits, and the exact config changes.

Drop your usage files here

Claude Code or Codex .jsonl, OpenCode msg_*.json, a Cursor usage CSV, or a ccusage report. Drop several at once. Read in this tab, never uploaded.

Choose files
Or in a terminal: npx tokenreceipt (finds the files for you)
Paste instead
Classic receipt: compare model by model

Make your receipt

Your data never leaves your browser
How do I export my usage?

How it works

  1. 01

    Export your usage

    Run `npx ccusage@latest daily --json`, copy Claude Code's /cost, or download a Cursor, Anthropic or OpenAI usage CSV.

  2. 02

    Paste or drop it here

    We detect the format and read it in your browser. Nothing is uploaded, and there's no account to make.

  3. 03

    Get your receipt

    See what the same tokens could cost on cheaper and free models, priced from a dated, sourced table.

  4. 04

    Share it

    Download a PNG in three sizes, copy a link or text, or post to X and LinkedIn. The link holds only your totals.

Private by design

Parsing runs on this page. The security policy blocks the page from sending your data anywhere. Analytics count page views and button clicks, never numbers.

Honest numbers

Every price has a source and a date (18 Oct 2026). Vendor pages win over resellers. We say "could cost about", because that's what an estimate is.

More formats

ccusage (daily, monthly, session, blocks), Claude Code /cost, Cursor, Anthropic and OpenAI CSVs, or manual entry. Unknown models are flagged, never priced at zero.

FAQ

Does my usage data get uploaded?

No. Your paste or file is read and priced by JavaScript on this page, in your browser. There is no upload endpoint, and the page's security policy only lets it talk to devprohacks.in and our page-view counter, which never receives usage numbers. A share link carries only the totals and model names shown on your receipt, never your raw logs. A report link from npx tokenreceipt keeps its summary after the #, which browsers never send to any server.

Is this what I would actually pay?

No, it's an estimate. We multiply your token counts by each model's public list price as of 18 Oct 2026. Real bills differ: other models use different tokenizers, cache differently, may need more or fewer tokens for the same task, and prices change. Every price is dated and sourced on /prices.

Is a cheaper model just as good?

Not necessarily. Cheaper isn't the same quality. A model that costs 95% less may need more retries, or may not handle your codebase as well. Use the receipt to decide what's worth testing, then test on your real work.

I'm on a Claude, ChatGPT or Cursor plan. What does this mean for me?

Plans charge a flat fee with usage limits, so you weren't billed per token. The receipt shows what the same tokens would cost at API list prices. That's useful to see the value you get from a plan, or what moving to pay-as-you-go on another model could cost.

Which exports are supported?

Raw logs from Claude Code, Codex CLI and OpenCode (drop the files, or run npx tokenreceipt), ccusage JSON (daily, weekly, monthly, session and blocks; covers Claude Code, Codex, OpenCode, Gemini CLI and more), Claude Code /cost output, Cursor's usage CSV, Anthropic Console usage CSV, OpenAI usage CSV, and manual entry. See /docs for exact commands.

Where do the prices come from?

Every day from two open price datasets, LiteLLM and models.dev (both MIT licensed), checked against each vendor's own pricing page. When they disagree, the vendor page wins and the row is flagged; big jumps are held for a human check. Every row has a source link and date on /index, and changes are listed there too.

Why is a free model on my receipt?

Some models are free for a limited time (for example GLM 5.3 Flash in OpenCode as of this week). We mark these with an asterisk: limits apply and the offer can end any time.

Do you use affiliate links?

No. We don't earn anything when you switch models, and we don't rank models by who pays us. We make money only from the optional Pro plan.