deeprelay benchmarks latest
Show the newest benchmark run
Show the newest benchmark run
Synopsis
Shows the newest published benchmark run. The table has one row per measured
lane: IN $/1M and OUT $/1M are the lane's list prices, TTFT p50 the
median time to first token and TOK/S p50 the median output rate over the
run's streamed trials, OK the trials that completed, and WEIGHTS the
weights-identity verdict (same_model for the deeprelay lane when it sits
inside the band the reference deployments define between themselves;
reference for the lanes that define the band). A (fingerprint) row per
model carries the overall verdict and its reason.
-o json prints the full report, including per-trial timings, the reference
band, per-lane distances, noise floors and the negative-control results.
Examples
deeprelay benchmarks latest
deeprelay benchmarks latest -o json | jq '.run.url'
deeprelay benchmarks latest -o json | jq '.models[] | {id, verdict: .fingerprint.verdict}'
deeprelay benchmarks latest [flags]
Options
-h, --help help for latest
Options inherited from parent commands
--api-base string API base URL (override with DEEPRELAY_API_BASE env) (default "https://api.deeprelay.ai/v1")
--debug Enable debug logging to stderr
--no-preflight Skip the plan/credit check before inference requests (override with DEEPRELAY_NO_PREFLIGHT env)
-o, --output string Output format: table|json (default table on TTY, json otherwise) (default "table")
SEE ALSO
- deeprelay benchmarks - Read the public inference benchmark (price, TTFT, throughput, weights check)