bench_run

{“what”:“Sequential numeric identifier, primary key for the run.”,“why”:“Provides a stable internal reference for joins and indexing.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:47.718454032+00:00”}

{“what”:“UUID assigned to the run, independent of the numeric id.”,“why”:“Allows external systems to reference the run without exposing the sequential id.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:47.861805601+00:00”}

{“what”:“Timestamp when the run was created, in UTC with timezone.”,“why”:“Needed to order runs and track when tests occurred.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:47.895064394+00:00”}

{“what”:“Benchmark mode, e.g., ‘load’ or ‘latency’.”,“why”:“Distinguishes different test types that affect metrics.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.166588182+00:00”}

{“what”:“Cloud provider or service name, e.g., ‘openai’.”,“why”:“Identifies which API provider was used for the run.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.343043886+00:00”}

{“what”:“Base URL of the API endpoint.”,“why”:“Required to reproduce the exact endpoint for reruns.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.673545908+00:00”}

{“what”:“Model name as sent in the API request.”,“why”:“Shows what the user asked for, independent of actual serving.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.710479714+00:00”}

{“what”:“Actual model name served by the provider, if different.”,“why”:“Captures model routing or aliasing that affects performance.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.771601727+00:00”}

{“what”:“Maximum context length in tokens supported by the model.”,“why”:“Determines feasible prompt sizes and influences test design.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.810614254+00:00”}

{“what”:“Maximum number of tokens the model can generate.”,“why”:“Limits output size and affects throughput calculations.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.845276470+00:00”}

{“what”:“Quantization level of the model, e.g., ‘int8’.”,“why”:“Affects performance and accuracy, needed for fair comparisons.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:48.923180022+00:00”}

{“what”:“Cloud region where the endpoint is hosted.”,“why”:“Important for latency analysis and cost estimation.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.056784298+00:00”}

{“what”:“Maximum tokens to generate per request.”,“why”:“Controls response length and test consistency.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.181813967+00:00”}

{“what”:“Desired number of tokens in the prompt.”,“why”:“Standardizes input size across runs for comparable metrics.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.246523611+00:00”}

{“what”:“Sampling temperature, typically 0.0 to 2.0.”,“why”:“Affects randomness of outputs; needed to reproduce results.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.435972969+00:00”}

{“what”:“Nucleus sampling probability mass, e.g., 0.9.”,“why”:“Another sampling parameter that influences output diversity.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.501572184+00:00”}

{“what”:“Level of reasoning effort, e.g., ‘low’, ‘high’.”,“why”:“Affects model behavior and latency, so must be recorded.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.531571799+00:00”}

{“what”:“Whether streaming responses were used (true/false).”,“why”:“Streaming changes latency measurement and resource usage.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.557670344+00:00”}

{“what”:“Random seed for reproducibility, if set.”,“why”:“Allows rerunning with identical randomness to verify results.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.586063931+00:00”}

{“what”:“Comma-separated list of concurrency levels tested, e.g., ‘1,10,50’.”,“why”:“Records load test parameters for scaling analysis.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.757178713+00:00”}

{“what”:“Hostname of the machine running the benchmark.”,“why”:“Identifies the test environment to detect hardware influence.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.776467732+00:00”}

{“what”:“Version of the benchrhythm tool used.”,“why”:“Ensures reproducibility across tool versions.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:04:49.814428542+00:00”}