Back to changelog

Usage Analytics by Model, Key and Member

See exactly where your spend goes. Every project gets a Cost by Model analytics page, every API key gets its own statistics page, and Enterprise orgs get per-member usage breakdowns — all on the date-range picker you already use. Member analytics are available on Enterprise.

Usage analytics on LLM Gateway: cost by model, per-API-key statistics, and per-member spend breakdowns

The dashboard could already tell you how much you spent and how many requests you sent — but not where it actually went. Which model is eating the budget? Which API key drives the traffic? Which person on the team? Analytics answers all three, with cost, request, and token breakdowns by model, by key, and by member.

Cost by model, on every project

Each project now has an Analytics page in the sidebar with two charts:

  • Cost by Model — a horizontal bar chart ranking your models, with Cost / Requests / Tokens tabs.
  • Cost by Model Over Time — a stacked area chart over your selected range, with the same tabs plus a Mappings / Canonical toggle that collapses provider-tagged variants (e.g. openai/gpt-5.5 and a custom mapping of the same model) into one canonical model.

Both derive from the same activity data the dashboard already reads, so they're timezone-correct, respect the shared date-range picker, and need no new data to populate.

Per-API-key statistics

Every API key gets a dedicated statistics page: summary cards for cost, tokens, requests, and error rate, plus the two cost-by-model charts scoped to that single key. The View Statistics action in the API-keys list now opens it directly, so you can see at a glance whether a key is healthy and what it's spending on.

Member analytics

Enterprise organizations also get usage broken down by person. The Members page adds a per-member table sorted by spend, and each member has a detail view:

Surface What it shows
Members table Per member: cost, tokens, requests, error rate, and API-key count.
Summary cards That member's cost, tokens, requests, and error rate for the range.
Most used Their top model, provider, and app.
Breakdowns Cost by model, plus top providers and top apps tables.

Usage is attributed by who created each API key, so spend lands on the member who owns the key. Member analytics are exposed through new GET /analytics/members and GET /analytics/members/{userId} endpoints, restricted to organization owners and admins on the Enterprise plan — non-enterprise orgs see an upgrade card, and non-admins see an access notice.


Cost breakdown docs → | Open your dashboard →