DailyReport (Claude Code) - 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 as backends of the Claude Code agent with Serper search and Jina Reader over MCP, extended thinking enabled; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.

Top models

#ModelScore
1GPT-5.4 (Medium)81.3
2GLM-576.7
3Claude Opus 4.6 (Thinking)75.6
4Kimi K2.571.8
5Gemini 3.1 Pro (Preview)68.4

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