Free LLM Cost Leak Calculator

Estimate how much of your OpenAI, Anthropic, Gemini, Bedrock, LangChain, gateway, voice, document, or coding-agent spend is likely waste. No login, no upload, no API key.

Your Usage Pattern

Use your best estimate if you do not have traces yet.

Monthly AI/API spend Main workload Average retry or failed-call rate 12% Requests using premium models by default 65% Reused context that could be cached 35% Prompt/history bloat 30% Tool-call or agent-loop amplification 3x Monthly growth in AI requests 18%

Get outside review

Estimated Leak Profile

Conservative screening estimate, not a bill audit.

Leak score 0 Enter usage assumptions to estimate waste.

Likely monthly waste $0

12-month run-rate waste $0

90-day savings target $0

Paid-review payback -

Turn this estimate into a paid cost read

Use the copied summary plus sanitized billing exports. No API keys, raw prompts, raw outputs, customer data, or credentials.

Pay $29 for rescue Pay $99 for triage Pay $299 for roast

Using Claude Code, Codex, Cursor, or OpenCode? Paste a status block into the usage checker.

After checkout, use the same email on the roast intake page and send the calculator summary, provider spend by model, retry/error counts, cache-token usage, and top workflows.

What to export

Provider or gateway spend by model, request volume, input/output/cache tokens, failures, retries, latency, and top workflows.

What not to send

No secrets, raw prompts, raw outputs, customer data, account IDs, credentials, or production write access.

When to pay

If the monthly waste estimate is larger than the fixed fee, a $29 rescue note, $99 triage, or $299 roast usually pays for itself quickly.