Helicone
Observability & evalsAn open-source observability platform that logs and analyzes LLM calls through a proxy.
Helicone observes model usage with minimal effort: route calls through its proxy — often a one-line change — and it logs requests, costs, latency, and errors, with dashboards and caching on top.
The proxy approach makes it one of the quickest ways to get visibility into model spend and behaviour, and it self-hosts.
Where it's ideally used
A fit when you want fast, low-effort visibility into model calls, costs, and latency through a simple proxy.
Where it doesn't fit
Proxy-based logging is less suited to deep, nested agent traces than a tracing-first tool.