LiteLLM proxy
The proxy's spend log: one row per request, with the model, the token counts, what the proxy charged, and the team and key it belongs to. Never the prompt or the response, which live in columns we do not ask for.
Access
- Kind
- A read-only key you paste
- What to give
- Proxy URL and an admin key
- Why that one
- LiteLLM runs on your infrastructure, so it asks for the address as well as the key. The key reads the spend log and nothing else.
What it attributes on its own
- From the source
- virtual key (hashed by the proxy) · team · end user (hashed) · model and upstream provider · cache hits · what the proxy charged
- Needs a rule or a trace
- outcome — a proxy records the call, not what it was for · workflow and agent unless a rule maps the team or the key
Connecting it
- Find your proxy's address. The URL your applications already call. It must be reachable from the internet for Valistry to read it.
- Use an admin key. The master key, or a key with admin rights. It is tested before it is stored. Open ↗
- Paste both and click Test and connect. Valistry reads the last seven days, reports how many requests and teams it found, then fills thirty.
Where the proxy fronts a provider you also connect directly, connect only one of the two: the same request would otherwise be counted twice, once by the proxy and once by the provider's own report.