Start free

Free, instant, works before signup

Find out what your team overspends

Run one command. Tokeven prices your last 30 days against the cheapest model that matches, and shows the cache savings you are leaving on the table.

Use Claude Code? Get your usage data with one command.

uvx tokeven teardown --open

Reads token counts and model IDs from your local Claude Code session logs, sends those numbers (never your prompts) to Tokeven, and opens your report here. Rather paste? Run uvx tokeven teardown and paste the output below.

Paste your Anthropic / OpenAI usage export, or just your monthly total

Accepts any of: a CSV export (model + token columns), a JSON array of per-call rows, a { "total_cost": 2400, "model_breakdown": {…} } object, or just a number like 2400. Rows we can’t identify are shown as unpriced, always flagged clearly.

fills the box so you can run it in one click

Your usage data is analyzed in memory and is not stored. For each run we keep the providers, models and top-line figures, plus a hashed (pseudonymous) email fingerprint if you gave one; our logs also note the company name you entered. Privacy Policy §2.7. Works before signup.

What the teardown checks

8

providers priced

88

models in the registry

$0

cost to find out