Tables

{“what”:“Quality metrics for a provider/model pair at a specific time.”,“why”:“Allows comparing model performance and tracking changes.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:59.504078872+00:00”}

{“what”:“Registry of AI inference providers with access, compliance, and rate-limit attributes.”,“why”:“Central catalog to manage provider credentials and routing decisions; without it, no single source of truth.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:24.053978873+00:00”}

{“what”:“List of approved model identifiers per provider, defining which models are permitted.”,“why”:“Without it, any model could be used, and there is no control over model availability.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:21.674696375+00:00”}

{“what”:“A single execution of a data collection job, with timing and outcome counts.”,“why”:“Without it, there is no record of when each collector ran and what it changed.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:22:41.234224977+00:00”}

{“what”:“Key-value store for DeepSeek configuration options.”,“why”:“Centralizes runtime settings so code can read them without hardcoding.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:22.322200493+00:00”}

{“what”:“Historical price records for offerings, one row per price observation.”,“why”:“Without it, past prices are unknown, so price changes cannot be audited or analyzed over time.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:49.543559945+00:00”}

{“what”:“Holds latency and throughput metrics for model inference benchmarks.”,“why”:“Without it, performance data is scattered and not comparable across providers.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T04:23:00.040423760+00:00”}

{“what”:“Aggregated inference performance metrics per provider/model over fixed time windows.”,“why”:“Provides a queryable history of latency, throughput, and success to detect regressions and capacity issues.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:14.772105857+00:00”}

{“what”:“One row per rate-limit tier, defining limits and costs for a provider or model.”,“why”:“Without it, rate limit information would be scattered or missing, preventing automated lookup of applicable limits.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:41.439889722+00:00”}

{“what”:“Registry of AI models with identifying attributes and metadata.”,“why”:“Without it, no canonical list of models exists for referencing in other tables.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:14:56.769460264+00:00”}