Ling-3.0-flash-Fin reportedly averages about 34% more output tokens per Intelligence Index task than VL
Artificial Analysis reports that Fin scored 967 Elo on the AA-Briefcase business-workflow benchmark, slightly below Ling-3.0-flash-VL’s 986. Fin scored slightly higher on presentation but lower on analytical quality.
TLDR
Artificial Analysis reports that Ling-3.0-flash-Fin averages about 67,000 output tokens per Intelligence Index task, versus about 50,000 for Ling-3.0-flash-VL—roughly 34% more.
On AA-Briefcase, which tests agents on complex business workflows using the available reporting to produce spreadsheets, presentations and memos, Artificial Analysis reports Fin scored 967 Elo against VL’s 986. Fin passed fewer rubric checks (23.5% versus 24.9%) and scored lower on Analytical Quality Elo (866 versus 907), but slightly higher on Presentation Elo (1,095 versus 1,076). Artificial Analysis notes that Fin lacks VL’s image-input capability.
Combined views
—
1 Source, first seen ago
