needaiforthis.Need AI For This
SponsorReelyze - See exactly why your Instagram Reels and TikToks underperform.
0
Inferock Bench logo

Inferock Bench

Prove exactly what you were billed for every LLM API call.

Quick answer

Inferock Bench — Prove exactly what you were billed for every LLM API call. It's freemium. Best for auditing LLM API billing to detect double-billed or miscounted token calls.

Inferock Bench is a source-available local proxy tool that sits between your application SDK and any OpenAI-, Anthropic-, or Gemini-compatible API, generating an independent per-call receipt of tokens consumed, failures encountered, and retries attempted. It is designed for engineering teams and platform operators running production LLM workloads who need verifiable, auditable records of their API usage rather than relying solely on provider billing dashboards. Integration requires only two lines of code, swapping the apiKey and baseURL in your existing SDK configuration, and your provider credentials are processed entirely on your local machine and never transmitted externally. This makes Inferock Bench particularly valuable for teams that have experienced or suspect billing discrepancies such as truncated responses, empty completions, double-billed calls, or miscounted token usage, giving them concrete evidence to dispute charges or diagnose cost anomalies. The project is released under the FSL-1.1-Apache-2.0 license, which converts automatically to the permissive Apache-2.0 license after two years, and the team publishes its methodology and accountability documentation openly rather than making unsubstantiated savings claims.

Key features

  • Per-call receipt generation capturing tokens, failures, and retries
  • Two-line SDK integration with no credential exposure outside your machine
  • Supports OpenAI, Anthropic, and Gemini API shapes via local proxy
  • Source-available codebase with published methodology and accountability pages

Pros & cons

PROS

  • +Credentials stay local and never leave your machine, protecting API key security
  • +Minimal integration effort with just two lines of configuration code required
  • +Transparent, source-available codebase with openly published methodology

CONS

  • The commercial Inferock service is invite-only and waitlisted, limiting access to its broader features
  • Service credits are gated on qualifying failures rather than offered as a blanket guarantee

Pricing

Free tier

Full Bench tool, source-available at no cost

Paid from

Invite-only, pricing on request

Enterprise

Invite-only commercial service, contact for access

Who is it for

  • Auditing LLM API billing to detect double-billed or miscounted token calls
  • Debugging production LLM traffic for truncated or empty API responses
  • Building internal cost accountability reports for teams using multiple LLM providers

Frequently asked questions

Is Inferock Bench free?

Yes, the Inferock Bench proxy tool is free to use. It is released under the FSL-1.1-Apache-2.0 license, which converts to the fully permissive Apache-2.0 license after two years. The separate commercial Inferock service is invite-only with no published pricing.

What is Inferock Bench best used for?

Inferock Bench is best used for auditing and verifying LLM API billing. It is ideal for teams that need independent, per-call receipts of token usage, failed requests, and retries to identify discrepancies like truncated responses, empty completions, or double-billed calls from providers like OpenAI, Anthropic, or Google Gemini.

What are the best alternatives to Inferock Bench?

Alternatives for LLM usage monitoring and cost tracking include LangSmith by LangChain, Helicone, Portkey, and OpenMeter. For general API observability, tools like Datadog or custom logging middleware can also serve similar audit purposes, though they lack Inferock Bench's specific focus on billing receipt generation.

Is Inferock Bench safe to use?

Yes, Inferock Bench is designed with security in mind. Your provider API keys are used locally only and never sent to Inferock or any third-party server. The codebase is source-available, so you can inspect exactly how it handles credentials and traffic before deploying it in your environment.

How much does Inferock Bench cost?

The Inferock Bench tool itself is free and source-available under FSL-1.1-Apache-2.0. The commercial Inferock service, which offers additional features and service credits, is currently invite-only and waitlisted with no publicly stated price. Interested teams should contact Inferock directly for commercial pricing details.

Reviews

No reviews yet. Be the first to review Inferock Bench.

Write a review

Links and HTML are removed. Be honest and respectful.

1
Raindrop Workshop logo

Debug AI agents locally with open source precision and speed.

0
Job Postings API logo

Unlock real-time access to 1.8M+ US job listings instantly.

0
Powabase logo

Build powerful AI apps with Postgres, RAG, and agents fast.

0
Respan Gateway logo

Route, observe, and evaluate every AI call in one place.

0
Proxon logo

Full visibility and control over every AI tool in your organization.

0
Warp logo

The AI-powered terminal that makes developers dramatically more productive.

Freemium