Arena, the crowdsourced model-evaluation platform formerly known as LMSYS/LMArena, has introduced factuality signals that measure whether models make true, verifiable claims — a shift that materially reorders how leading models rank against one another. The addition, unveiled on July 15, 2026, blends traditional human preference scores with an objective measure of factual accuracy across both the Text Arena and Search Arena leaderboards.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.