Labs / Z.ai
Does the standard hold across Z.ai's models?
Every model the Benchmark tests from Z.ai, checked against one line: with guidance, fewer than 10 in 100 drafts needing revision. The Benchmark tests one model from this lab, so there is no family comparison yet.
Human-centered language, with guidance
68 drafts per model · bar shows the 95% interval · dashed line at 10%Holds: the whole interval sits under the line. Falls short: the whole interval sits above it. Can't tell yet: the interval crosses the line, and more drafts are needed to say.
Trauma-informed language, with guidance
24 drafts per model · bar shows the 95% interval · dashed line at 10% GLM 5.3 Settled
Holds: the whole interval sits under the line. Falls short: the whole interval sits above it. Can't tell yet: the interval crosses the line, and more drafts are needed to say.
Generations in this lab
No same-run pair this month.
New models, in detailZ.ai models the Benchmark doesn't test
This list comes from the lineage registry, which isn't built yet. It will name each untested model and why, so silence never reads as a pass.