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.