Helicone
Apache-2.0 gateway in Rust, with observability built in rather than bolted on.
88
sovereigntyHelicone is an open-source AI gateway written in Rust, with logging, caching and cost tracking built into the routing layer rather than added alongside it — which is the practical difference from LiteLLM, where observability is thin and usually paired with something else. It is Apache-2.0 throughout, self-hostable, and also offered as a hosted service with a free tier. For teams who want one component instead of two, it is the tidier answer.
OPEN SOURCEApache-2.0SELF-HOSTLOCAL-FIRST
LicenseApache-2.0
PricingFree and self-hostable under Apache-2.0. The hosted service has a free tier of roughly 10,000 requests per month, with paid plans starting around $79/month.
Open sourceYes
Self-hostableYes
Local-first dataYes
What it does well
- +Apache-2.0 with no enterprise-directory carve-out
- +Routing and observability in one component — less to assemble
- +Written in Rust; low latency overhead for a proxy
- +Hosted option with a real free tier if you would rather not run it
Where it falls short
- −Smaller project and community than LiteLLM (about 6k stars)
- −Fewer providers supported than LiteLLM's 100+
- −The hosted tier means prompts transit a third party — the thing self-hosting avoided
- −Less mature budget and virtual-key tooling than LiteLLM
Helicone as an alternative to
Where Helicone shows up in our comparisons, and how it ranked.
Helicone head-to-head
Straight comparisons against the tools people weigh it against.