Not All Needles Are Found: How Fact Distribution and Don't Make It Up Prompts Shape Literal Extraction, Logical Inference, and Hallucination Risks in Long-Context LLMs
This addresses reliability issues for enterprise and research users deploying LLMs with large document inputs, but it is incremental as it builds on existing needle-in-a-haystack benchmarks.
The study investigated how fact placement, corpus-level distributions, and anti-hallucination prompts affect long-context LLMs' performance in literal extraction, logical inference, and hallucination risks, finding that longer contexts can degrade performance when evidence is diluted, with models showing varied robustness and anti-hallucination instructions sometimes reducing accuracy.
Large language models (LLMs) increasingly support very long input contexts. Yet it remains unclear how reliably they extract and infer information at scale. Performance varies with context length and strongly interacts with how information is distributed in real-world corpora. Motivated by these observations, we study how fact placement, corpus-level fact distributions, and Don't Make It Up prompts influence model behavior. We introduce an extended needle-in-a-haystack benchmark across four production-scale models: Gemini-2.5-flash, ChatGPT-5-mini, Claude-4.5-haiku, and Deepseek-v3.2-chat. Unlike prior work, we separately evaluate literal extraction, logical inference, and hallucination risk. Our study considers both positional effects and realistic distributions of evidence across long contexts, as well as prompts that explicitly discourage fabrication. We find that longer contexts alone do not guarantee better performance and can be detrimental when relevant evidence is diluted or widely dispersed. Performance varies substantially across models: some show severe degradation under realistic conditions, while others remain more robust at longer context lengths. Anti-hallucination (AH) instructions can make some models overly conservative, sharply reducing accuracy in literal extraction and logical inference. While we do not directly compare retrieval-augmented generation (RAG) and cache-augmented generation (CAG), our results suggest many failures stem from ineffective context utilization. Models often struggle to identify and prioritize relevant information even when it is present. These findings have direct practical implications, as enterprise workflows increasingly involve pasting large volumes of unfiltered documents into LLM prompts. Effective context length and model-specific robustness to long contexts are therefore critical for reliable LLM deployment in research and business.