Hanzo
OpenapiBenchmark

Returns each model's measured score per run over time, oldest first, with the…

Returns each model's measured score per run over time, oldest first, with the change between runs.

GET /v1/benchmark/history

Addresshttps://api.hanzo.ai/v1/benchmark/history
MethodGET
Operationget_benchmark_history
AuthAuthorization: Bearer $HANZO_API_KEY

Returns each model's measured score per run over time, oldest first, with the change between runs.

This is the counterweight to a leaderboard: the board shows the latest run because that is what "how good is it" means, and a single latest number cannot distinguish a model that has always been strong from one that just improved, or from one that regressed after a provider changed something. Both matter for routing, and only one of them is visible on a board.

Runs with no id — attempts recorded before runs existed — group under the empty run, which is honestly what they are: one undated measurement.

Request

2 fields.

FieldInTypeRequiredDescription
BenchmarkquerystringBenchmark is the catalog id to read, defaulting to gpqa_diamond.
ModelquerystringModel filters to one model.

Response

StatusBodyMeaning
200historyOutok

200 body — 11 fields.

FieldInTypeAlwaysDescription
benchmarkbodystringBenchmark is the catalog id these histories are about.
databodyModelHistory[]Data is one entry per model, ordered by model name.
data[].modelbodystringModel is the system these runs measured.
data[].pointsbodyRunPoint[]Points is every run, oldest first.
data[].points[].atbodystring (date-time)At is when the run was recorded.
data[].points[].deltabodynumberDelta is the change in score from the previous run for this model, absent on the first.
data[].points[].nbodyintegerN is how many items the run covered.
data[].points[].runbodystringRun is the measurement id these attempts were recorded under.
data[].points[].scorebodynumberScore is accuracy over the items this run covered, as a percentage.
data[].trendbodynumberTrend is the change from the first run to the last, absent when there has only been one.
totalbodyintegerTotal is how many models Data holds.

Failure carries the platform error shape — see Errors.

Examples

hanzo benchmark history

Benchmark API · All Hanzo APIs · Interactive reference

How is this guide?

On this page