PRISM Leaderboard

Playwright

PRISM Benchmark

Quarter	Version	RPS-Index	R̄	S̄	Ī	False Heals
2026-Q4	1.60.0	0.4599	0.6409	0.6900	0.5200	8 of 50

Entries labelled operator baseline were run by the benchmark’s maintainers. They are operator-reported and are not third-party reproducible.
The published evaluator will not run end to end: there is no public sandbox, and there is not going to be one. Runs happen on operator infrastructure, quarterly.
This board is small and mostly operator-run. That is the honest state of it, not a staging artefact.

How this is measured · The benchmark on GitHub