Helicone

Helicone

By Helicone

Open-source LLM observability via a one-line proxy — logging, caching, and cost tracking

No updates in 6+ monthsv2025.08.21-12025-08-21TypeScriptApache-2.0
6.0k Open source Free tier Paid

Doc version 1.0 · Updated

Related concepts

Related tools

Overview

Helicone is an open-source observability platform for LLM apps that works as a proxy: route your provider calls through its gateway and get logging, caching, rate limiting, and cost tracking with a one-line change.
It runs as managed Helicone Cloud or self-hosted, and also offers an async SDK when you would rather not proxy.

flowchart LR
  app[your app] --> h[(Helicone proxy)]
  h --> provider[LLM provider]
  h --> dash[logs / cost / cache]

The Code samples tab shows the proxy approach with the OpenAI SDK.

When to use it

Choose Helicone when you want the fastest path to logging and cost visibility — a gateway you drop in front of any OpenAI-compatible provider — with self-hosting available for data control.