Native-Benchmark Challenge

The strictest room in the hall: dense read-alouds with native-reference scoring on every clip (dual-locale Accent Gap on each), C1+ one-play listening and hard dictation. Benchmarked against native-speaker reference models — explicitly, honestly, as a stretch target. Three attempts included — every attempt draws a fresh set. Scored by AssessAll's AI examiner panel: accent-neutral grading for content, grammar and fluency, with market-listener intelligibility surfaced as a coaching lens.