---
title: Free AI Waste Check — Anonymous Usage Cost Scan
description: Upload your OpenAI or Anthropic usage CSV and instantly see where your AI spend is going. Costs are recomputed from a verified model-price registry. Anonymous, parsed in memory, zero database writes, deleted after 30 minutes.
canonical: https://cromus.ai/waste-check
source_html: https://cromus.ai/waste-check
---

# Free AI Waste Check

> Upload an OpenAI or Anthropic usage export and instantly see where your AI spend is going — with every cost recomputed from a verified model-price registry. No account required, nothing saved.

---

## What it does

The AI Waste Check analyzes an AI usage export (OpenAI or Anthropic CSV) and returns:

- **Recomputed total spend** — every cost is recalculated from Cromus's verified model-price registry, never trusting the numbers in your file.
- **Spend by model** — a per-model breakdown showing where the money goes.
- **Findings** — plain-language signals such as model concentration, unmapped models, output-heavy usage, and cache usage.
- **Savings scenarios** — registry-grounded model-substitution estimates (upper bound; validate quality before switching).

## How your data is handled

The Waste Check is privacy-first by design:

- Your file is **security-scanned before parsing**. Files over 10MB, or files that fail the scan, are rejected.
- Parsing happens **entirely in memory**. The raw file is **discarded immediately** after parsing — never written to disk or to our database.
- Only an **aggregate summary** is computed — never your row-level data.
- The summary lives in a **temporary in-memory store for 30 minutes**, then is deleted automatically. **Nothing is written to our database.**
- Your usage data is **never sent to an LLM or any third party**. The analysis is deterministic local computation — **zero AI calls**.
- The upload and analysis endpoints are **excluded from request logging**, so file contents never appear in server logs.
- Optional **share links** show **rounded summary figures only** — no row-level data, identifiers, or filenames.

## After the check

No account is required to run a Waste Check. To track waste over time, sign in to:

- Map imported rows to specific workflows.
- Measure **Croms™** — units of preventable AI workflow waste.
- Run the **Reality Check loop** (Predict → Execute → Log → Learn) to calibrate predictions against reality.

Run a free Waste Check: https://cromus.ai/waste-check