DailyReport (Search Tools) - Factuality: leaderboard

Metric: Factuality (0-1 scaled to %), the share of extracted objective claims verified correct by web search, counted only where the instruction-following rubric holds; open-ended daily search tasks on trending topics decomposed into subtasks with cascade rubrics, judged by Gemini 3 Flash; LLMs with google_search and fetch_webpage tools, extended thinking or reasoning enabled; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.

Top models

#ModelScore
1GPT-5.4 (Medium)83.5
2Claude Opus 4.6 (Thinking)79.6
3GLM-578.4
4Gemini 3.1 Pro (Preview)73
5Kimi K2.572.8

Interactive version: theaggregate.ai/benchmark?slug=dailyreport-search-tools-factuality · How It Works · Data refreshed daily, snapshot 2026-09-29.