latency_benchmarks

{“what”:“Unique integer identifier for each benchmark row.”,“why”:“Without it, rows cannot be referenced or updated individually.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.112924536+00:00”}

{“what”:“Identifier of the service provider (e.g., ‘openai’, ‘anthropic’).”,“why”:“Groups benchmarks by provider to compare performance across vendors.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.153824032+00:00”}

{“what”:“Identifier of the model version (e.g., ‘gpt-4o’).”,“why”:“Distinguishes benchmarks for different models from the same provider.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.176926213+00:00”}

{“what”:“Origin of the benchmark data, default ‘openrhythm’.”,“why”:“Tracks which tool or pipeline produced the measurement.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.209358848+00:00”}

{“what”:“Average time to first token in milliseconds.”,“why”:“Quantifies perceived responsiveness; missing it hides startup latency.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.234465045+00:00”}

{“what”:“Median time to first token in milliseconds.”,“why”:“Shows typical latency unaffected by outliers.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.250400707+00:00”}

{“what”:“95th percentile time to first token in milliseconds.”,“why”:“Reveals worst-case latency for tail latency analysis.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.279703103+00:00”}

{“what”:“Average tokens generated per second.”,“why”:“Measures throughput; without it, speed comparisons are impossible.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.302077902+00:00”}

{“what”:“Average end-to-end request latency in milliseconds.”,“why”:“Captures total response time including network and processing.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.314779109+00:00”}

{“what”:“Number of requests used for the benchmark.”,“why”:“Indicates statistical confidence; low counts make averages unreliable.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.359441716+00:00”}

{“what”:“Fraction of successful requests (0.0 to 1.0).”,“why”:“Reveals reliability; missing it hides request failures.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.371876526+00:00”}

{“what”:“Timestamp when the benchmark was recorded.”,“why”:“Enables time-based comparisons and detects performance drift.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.394589571+00:00”}