Ling-3.0-flash-VL has a 22% hallucination rate, ArtificialAnlys reports
ArtificialAnlys says the model attempted around 34% of AA-Omniscience questions and produced roughly 50,000 output tokens per Intelligence Index task.
TLDR
ArtificialAnlys describes Ling-3.0-flash-VL’s 22% hallucination rate as “relatively low,” while noting that it attempted around 34% of AA-Omniscience questions. The account also reports that the model produced roughly 50,000 output tokens per Intelligence Index task.
Combined views
—
1 Source, first seen ago
