Zero install · runs in your browser

Which files are burning your
Claude / GPT tokens?

Paste your code below. See the token cost and dollar spend — instantly, before you run a single API call.

0.00%
estimation error
4
LLM providers
42.9%
avg compaction saving
1.8M
tokens/sec scan
Paste code or file contents 0 chars
Analysis

Paste code on the left and click Analyze to see token cost breakdown

Why token costs are quadratic — not linear

Each LLM re-reads the entire conversation history on every turn. A 20-turn session can cost 120× more than turn 1.

Run it on your actual codebase

The CLI scans your full repo — all files, real token counts, CI budget gates.

pip install -e ".[tiktoken]" && distill scan --path .