GET
/
v0
/
endpoint-metrics

Extracts cost and speed data for the provided endpoint via our standardized efficiency benchmarks, in the specified region, with the specified sequence length, with all benchmark values returned within the specified time window.

When extracting data for a custom endpoint, then model is the endpoint name, and provider must be set as "custom". The arguments region and seq_len are ignored for custom endpoints (they are not publishable).

If neither start_time nor end_time are provided, then only the latest benchmark data is returned. If only start_time is provided, then end_time is assumed to be the current time. An exception is raised if only end_time is provided.

Authorizations

Authorization
string
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Query Parameters

model
string

Name of the model.

provider
string

Name of the provider.

region
string
default: "Iowa"

Region where the benchmark is run. Options are: "Belgium", "Hong Kong" or "Iowa".

seq_len
string
default: "short"

Length of the sequence used for benchmarking, can be short or long

start_time
string

Window start time. Only returns the latest benchmark if unspecified

end_time
string

Window end time. Assumed to be the current time if this is unspecified and start_time is specified. Only the latest benchmark is returned if both are unspecified.