ModelCensusfailure-mode benchmark
Studio

claude-haiku-4.5.

This loop has not been published. Its metadata is shown; its numbers are not.

loop 2026-08-09 · claude-haiku-4.5 as served · set s1.0 · cases v1.2 · conditions v0.9-no-control · ground truth as of 2026-08-09

Draft

Execution is incomplete or validation has not passed. Nothing measured here is final. Rates, transcripts, and verdicts appear only after publish — a figure taken from a draft travels without the caveat that it was going to move.

1 cards · 1 with a residual written

Back to Studio