LLM spend by group and model
Even-split API-equivalent spend per group × model — an estimate, not billed spend.
Even-split API-equivalent spend per group × model — an estimate, not billed spend.
ApiKeyAuthAuthorizationBearer <token>Bearer token authentication. Send an API key (a saai_api_-prefixed token) as Authorization: Bearer <token>.
Required permission
view_user_analyticsscope?stringown returns only the caller's own records, even when they may see the whole organization. all, the default, returns everything the caller may see. It never widens what the caller may see.
"all""own"start_time*stringStart of the time range to report on, inclusive.
date-timeend_time*stringEnd of the time range to report on, exclusive.
date-timetelemetry_provider?array<>Filter to OpenTelemetry providers; LLM gateway calls have none, so any value excludes them.
agent_slug?array<string>Filter to agent types by catalog slug, such as claude-code.
agent_instance_id?array<>Filter to specific agent instances.
model_provider_id?array<>Filter to model providers: the inference connections that served the calls.
model?array<string>Filter to specific model names.
user_group?array<string>Filter to users in specific groups. The Unassigned ("") bucket cannot be filtered to. Narrows the even split as well as the rows, so only the selected groups come back.
auth_mode?array<>The request has succeeded.
application/json- response
Group × model spend matrix.
items*array<>curl -X GET 'https://your-org.aisecurity.services.us.connect.secureauth.com/api/v1/analytics/llm-spend-by-group-model?start_time=2019-08-24T14%3A15%3A22Z&end_time=2019-08-24T14%3A15%3A22Z'{ "items": [ { "group": "string", "model": "string", "cost": 0.1 } ]}List LLM sessions GET
Sessions ranked by cost or tokens, with per-session usage.
LLM spend summary GET
Subscription-vs-API cost split for a time range, using configured seat pricing. Accepts only seat-coherent filters: telemetry_provider, user, and group scope both the usage legs and the priced seats; a model can't scope a seat, so there is no model param (and no auth_mode — the endpoint decomposes by auth path itself).