Inferock Bench

An independent receipt for every LLM API call

Visit website →

About

Inferock Bench is a local proxy that sits between your app and OpenAI, Anthropic, Gemini, or OpenRouter-shaped calls. It captures per-call token usage, failures, and retries, then generates an independent receipt showing what you were actually billed and how much you may be overpaying without realizing it. Point your existing SDK at it by changing two settings, apiKey and baseURL, and it captures every call as an independent, auditable record.

The tool is built on a simple observation: the company charging you also decides what counts as a failure, and it alone keeps the detailed records. Inferock Bench catches answers cut off mid-stream that still got billed in full, empty replies with tokens attached, token counts that don't match visible output, retries that may have silently doubled a charge, and cache discounts you may be missing on your invoice. Every run reports spend observed, bill-bound money loss, and time loss, with a separate invoice-check exposure line that is never added into the money-loss total.

Your provider key never touches Inferock's servers; it's used locally only, attached directly to provider requests. It's open source under FSL-1.1-Apache-2.0, converting automatically to full Apache-2.0 in two years. It runs in about a minute via a single npx command.