benchmarks.site

Public leaderboard · Hallucination & factuality

AA-Omniscience

Tests how much factual knowledge an AI really has across 42 topics, and punishes it for confidently making things up instead of saying 'I don't know'. A score of 0 means as many wrong as right answers.

Leaderboard

Top 3 right now

  1. 1stClaude Fable 5.2 (Max, Default Fallback)67
  2. 2ndClaude Fable 5.2 (Xhigh, Default Fallback)65
  3. 3rdClaude Fable 5.2 (High, Default Fallback)64

What it measures

6,000 knowledge questions across 42 topics, scored with an index from -100 to 100 that penalizes wrong answers and rewards abstaining when unsure.

How to read the numbers

Omniscience Index (-100 to 100): rewards correct answers, penalizes wrong answers, neutral for abstaining; higher is better — higher is better.

Results

3 rows · higher is better (Omniscience Index)CSV
1Claude Fable 5.2 (Max, Default Fallback)6779
2Claude Fable 5.2 (Xhigh, Default Fallback)6578
3Claude Fable 5.2 (High, Default Fallback)6477

Note: Only the top 3 are shown in the page's rendered text; the full table is client-side and the embedded data could not be reliably matched to model names, so rows are limited to these 3 (values rounded as displayed). Benchmark v1.1 (Oct 2026).

Source: Artificial Analysis · artificialanalysis.ai · retrieved 2026-11-17

Markdown · More Hallucination & factuality