Perplexity's Decider V1.2 Tops Vals AI's New ClaimBench

Vals AI launched ClaimBench, a benchmark evaluating 15 decision models on 1,960 financial claims to judge whether documents support, contradict, or don't address each claim. Perplexity's newly released pplx-decider-v1.2 topped the leaderboard.

2026-10-21 ~ 2026-10-21 · 3 related posts