bench


Description

{“what”:“One row per LLM inference request in a benchmark run.”,“why”:“Without it, per-request performance and cost data would be lost.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.711478241+00:00”}

Columns

Column Type Size Nulls Auto Default Children Parents Comments
id bigserial 19 √ nextval('bench_id_seq'::regclass)

{“what”:“Auto-incrementing bigint primary key for each row.”,“why”:“Without it, rows cannot be uniquely referenced or updated.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.722854597+00:00”}

run_id uuid 2147483647 null

{“what”:“UUID identifying the benchmark run this row belongs to.”,“why”:“Without it, rows from different runs cannot be grouped.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.740927679+00:00”}

ts timestamptz 35,6 now()

{“what”:“UTC timestamp of request start; defaults to now().”,“why”:“Without it, time-based analysis of latency or load is impossible.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.752022842+00:00”}

provider text 2147483647 null

{“what”:“Name of the LLM provider (e.g., OpenAI, Anthropic).”,“why”:“Without it, cost and performance cannot be attributed to a vendor.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.766700411+00:00”}

route text 2147483647 null

{“what”:“API endpoint path used (e.g., /v1/chat/completions).”,“why”:“Without it, different API versions or endpoints cannot be distinguished.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.781387315+00:00”}

model text 2147483647 null

{“what”:“Model identifier string (e.g., gpt-4o, claude-3.5-sonnet).”,“why”:“Without it, results cannot be compared across models.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.792828271+00:00”}

iteration int4 10 null

{“what”:“Zero-based index of this request within its run.”,“why”:“Without it, repeated attempts for the same prompt cannot be ordered.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.804198816+00:00”}

prompt_label text 2147483647 null

{“what”:“Human-readable label for the prompt used (e.g., ‘math-1’).”,“why”:“Without it, results cannot be tied to a specific test case.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.820974483+00:00”}

max_tokens int4 10 null

{“what”:“Maximum number of tokens allowed in the completion.”,“why”:“Without it, token usage and cost cannot be normalized.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.845241093+00:00”}

temperature float8 17,17 null

{“what”:“Sampling temperature value (0.0 to 1.0) used for generation.”,“why”:“Without it, output variability cannot be explained or controlled.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.865531553+00:00”}

seed int8 19 √ null

{“what”:“Optional random seed for deterministic generation; null if unset.”,“why”:“Without it, reproducibility of outputs cannot be verified.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.883615505+00:00”}

stream bool 1 null

{“what”:“Boolean indicating if response was streamed (true) or buffered (false).”,“why”:“Without it, timing metrics like ttft are not comparable.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.897821750+00:00”}

prompt_tokens int8 19 0

{“what”:“Number of tokens in the input prompt; defaults to 0.”,“why”:“Without it, cost and throughput calculations are incomplete.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.915187422+00:00”}

completion_tokens int8 19 0

{“what”:“Number of tokens in the generated output; defaults to 0.”,“why”:“Without it, output length and cost cannot be measured.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.929619939+00:00”}

ttft_ms int8 19 √ null

{“what”:“Time to first token in milliseconds; null if not measured.”,“why”:“Without it, perceived latency of streaming responses is unknown.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.947477583+00:00”}

total_ms int8 19 null

{“what”:“Total request duration in milliseconds from start to finish.”,“why”:“Without it, end-to-end latency cannot be compared.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.962216329+00:00”}

output_tok_s float8 17,17 √ null

{“what”:“Output tokens per second, computed from completion_tokens and timing.”,“why”:“Without it, generation speed cannot be compared across models.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.976672521+00:00”}

cost_usd float8 17,17 0

{“what”:“Cost of the request in US dollars, computed from token counts and pricing.”,“why”:“Without it, total spend per run or model cannot be calculated.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:55.996067223+00:00”}

status int4 10 null

{“what”:“HTTP status code or custom status code of the response (e.g., 200, 429).”,“why”:“Without it, failed or rate-limited requests cannot be identified.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:56.013969329+00:00”}

error text 2147483647 √ null

{“what”:“Error message or stack trace if the request failed; null on success.”,“why”:“Without it, failure reasons are unknown, hindering debugging.”,“by”:“fleet-annotator-worker”,“at”:“2026-09-23T00:01:56.038162477+00:00”}

Indexes

Constraint Name Type Sort Column(s)
bench_pkey Primary key Asc id
bench_provider_idx Performance Asc/Asc provider + completion_tokens
bench_run_idx Performance Asc/Asc/Asc run_id + provider + iteration
bench_ts_idx Performance Desc ts

Relationships